City University of New York (CUNY) CUNY Academic Works
All Dissertations, Theses, and Capstone Projects Dissertations, Theses, and Capstone Projects
6-2020
The Functional Role of the Anterior Insular Cortex in Cognitive Control
Yu Chen The Graduate Center, City University of New York
How does access to this work benefit ou?y Let us know!
More information about this work at: https://academicworks.cuny.edu/gc_etds/3705 Discover additional works at: https://academicworks.cuny.edu
This work is made publicly available by the City University of New York (CUNY). Contact: [email protected] THE FUNCTIONAL ROLE OF THE ANTERIOR INSULAR CORTEX IN COGNITIVE
CONTROL
by
YU CHEN
A dissertation submitted to the Graduate Faculty in Psychology in partial fulfillment of the
requirements for the degree of Doctor of Philosophy, The City University of New York
2020
© 2020
YU CHEN
All Rights Reserved
ii
The Functional Role of The Anterior Insular Cortex in Cognitive Control
by
Yu Chen
This manuscript has been read and accepted for the Graduate Faculty in Psychology in
satisfaction of the dissertation requirement for the degree of Doctor of Philosophy.
Date Jin Fan, Ph.D.
Chair of Examining Committee
Date Richard Bodnar, Ph.D.
Executive Officer
Jin Fan, Ph.D.
Jeff Beeler, Ph.D.
Elizabeth Chua, Ph.D.
Tatiana Emmanouil, Ph.D.
Supervisory Committee
THE CITY UNIVERSITY OF NEW YORK
iii
ABSTRACT
The Functional Role of the Anterior Insular Cortex in Cognitive Control
by
Yu Chen
Advisors: Jin Fan, Ph. D. & Jeff Beeler, Ph. D.
Cognitive control, a higher level psychological construct, refers to efficient coordination of thoughts and actions for the accomplishment of goal-directed behaviors. Cognitive control is supported by a commonly activated cognitive control network, and the anterior insular cortex
(AIC) serves as one of its key structures. However, the functional role of the AIC in cognitive control has not been fully understood. A human lesion study was conducted to examine the necessary function of the AIC in cognitive control. A mouse optogenetic study with fiber photometry recording further examined whether the bilateral AIC was important for cognitive control and how the AIC played a role in different stages of cognitive control (e.g., state uncertainty processing, execution of control, or motor generation). Compatible versions of the post-target interference task consisting of congruent and incongruent conditions were used to measure cognitive control in humans and mice, respectively. In the human lesion study, the patients with lesions in the AIC showed longer overall response time (RT), lower overall processing efficiency, and greater conflict effects of RT and processing efficiency. These findings provided lesion-based evidence to support a causally necessary function of the AIC in cognitive control. In the mouse study, the accuracy of the congruent condition decreased when the AIC was silenced unilaterally or bilaterally by optogenetics after the cue sound and when the
AIC was silenced bilaterally during the presentation of target and distractor stimuli, indicating that the disruption of the AIC resulted in a reduction in global processing efficiency. The fiber
iv
photometry results showed a significant decrease of the calcium-dependent signal after the cue sound compared to baseline, suggesting that the AIC was involved in state uncertainty processing. The results of the human lesion study identified the necessary role of the AIC in cognitive control. The findings of the mouse study further demonstrated the role of the AIC in cognitive control in both hemispheres and suggested a critical role of the AIC in state uncertainty processing.
Keywords: anterior insular cortex, cognitive control, lesion, mouse model, optogenetics, fiber photometry
v
CONTENTS
Abstract iv
Contents vi
Tables x
Figures xi
Chapter 1: Research Objective 1
Introduction 1
Rationale for study 1
Research questions and hypothesis 2
Chapter 2: Literature Review 4
Definition of cognitive control 4
Cognitive control in human studies 5
Measurement of cognitive control 5
Cognitive control network 6
Capacity of cognitive control 6
Processing efficiency of cognitive control 7
Top-down and bottom-up cognitive control 8
The insular cortex and its anterior part 9
Structural and functional connectivity of the AIC 10
Lesions in the AIC and cognitive control 11
Cognitive control in mouse studies 14
The mouse as a model mechanism 14
Measurement of cognitive control 16
vi
Neural substrates underlying cognitive control 17
Structural and functional features of insular cortex 18
Context for the proposed study 20
Chapter 3: Anterior Insular Cortex is Necessary for Cognitive Control:
A Human Lesion Study 21
Abstract 21
Introduction 22
Methods 24
Participants 24
Lesion reconstruction 26
Post-target interference task 27
Data analysis 29
Results 31
Comparisons between the AIC, NC, and BDC groups 31
The mean RT, error rate, and efficiency 31
Conflict effects 31
Oddball effects 33
Comparisons between the ACC, NC, and BDC groups 33
Discussion 33
A necessary role of the AIC in the processing efficiency of cognitive control 33
Distinctions between the roles of the AIC and the ACC in cognitive control 36
Top-down and bottom-up cognitive control 39
Conclusion 40
vii
Chapter 4: Anterior Insular Cortex is Critical for State Uncertainty Processing:
A Mouse Study 41
Abstract 41
Introduction 42
Methods 44
Animals 44
Chamber setup 45
Training protocol 46
Post-target interference task and paradigm validation 49
Virus 50
Surgery protocol 50
Optogenetics setup 52
Behavioral testing with optogenetic inhibition 53
Fiber photometry 54
Histology 56
Data analysis 56
Paradigm validation 56
Optogenetic inhibition 57
Fiber photometry recording 58
Results 59
Paradigm validation 59
Optogenetic inhibition 60
Results of the experimental group 62
viii
Results of the control group 63
Fiber photometry 65
Target-locked results 65
Response-locked results 69
Reward-related processing 71
Discussion 72
The role of the AIC in cognitive control in the mouse model 72
State uncertainty processing 72
Network global efficiency 73
Reward-based association learning 74
A compensatory role of the hemispheric AIC 76
Neuroanatomy of the AIC 76
Conclusion 77
Chapter 5: General Discussion 78
Cognitive control: from mouse to human 78
The AIC, the processing efficiency of cognitive control, and the CCC 78
The AIC and uncertainty processing 79
The AIC and reward-based association learning 80
The AIC, cognitive control, and higher level cognition 81
A functional architecture of cognitive control 81
Conclusion 82
References 84
ix
TABLES
Table 1. Participant characteristics in the human lesion study
Table 2. Parameters of the paradigm in the training sessions for mice
Table 3. Mean and standard deviation (SD) of the number of total trials, and omission rate, outlier rate, accuracy, and RT for all conditions (overall), congruent condition (cong), and incongruent condition (incong)
Table 4. Mean (SD) of the overall omission rate and the overall outlier rate for different experiments and groups
Table 5. The mean and SD of activation amplitude, activation duration, inhibition amplitude, and inhibition duration in the event windows of cue-to-lever, lever-to-target, target presentation, and distractor presentation for the correct and incorrect trials
Table 6. The mean and SD of activation amplitude, activation duration, inhibition amplitude, and inhibition duration in the event windows of 0 to 4 s and 4 to 8 s after response for correct and incorrect trials
x
FIGURES
Figure 1. Lesion mapping for patients with unilateral lesions in the anterior insular cortex (AIC group) and in the anterior cingulate cortex (ACC group)
Figure 2. Schematic of the post-target interference task in the human study
Figure 3. Behavioral performance of post-target interference task in the NC, BDC, AIC, and
ACC groups
Figure 4. Chamber setup and feeding periods in the training and testing sessions
Figure 5. Schematic of the post-target interference task in the mouse study
Figure 6. Optogenetic inhibition in the post-target interference task
Figure 7. Timeline for the analysis in the fiber photometry recording
Figure 8. Viral expression and sites of ferrule placement in all experiments
Figure 9. Accuracy of conditions in the experimental group with/without unilateral inhibition on the AIC
Figure 10. Accuracy of conditions in the experimental group with/without bilateral inhibition on the AIC
Figure 11. Accuracy of conditions in the control group with/without unilateral inhibition on the
AIC
Figure 12. Accuracy of conditions in the control group with/without bilateral inhibition on the
AIC
Figure 13. Target-locked averaged calcium transient in response to events in correct trials (black line) and incorrect trials (red line) across time
Figure 14. Target-locked averaged calcium transient in response to events in the congruent condition (solid line) and the incongruent condition (dashed line) in correct trials across time
xi
Figure 15. Response-locked averaged calcium transient in correct trials (black line) and incorrect trials (red line) across time.
Figure 16. Response-locked averaged calcium transient in response to events in the congruent condition (solid line) and the incongruent condition (dashed line) in the correct and incorrect trials across time.
Figure 17. Response-locked averaged calcium transient in correct-reward trials with delayed reward interval of 200 ms (blue), 500 ms (yellow), and 800 ms (magenta), correct-no-reward trials (green), and incorrect trials (red) across time.
Figure 18. Delivery-locked averaged calcium transient in all correct-reward trials.
xii
CHAPTER 1
Research Objective
Introduction
Cognitive control, a high-level psychological construct, refers to the ability to flexibly coordinate thoughts and actions for the accomplishment of goal-directed behaviors (Fan, 2014;
Mackie, Van Dam, & Fan, 2013). With a limited capacity of 3 to 4 bits per second in information processing (T. Wu, Dufford, Mackie, Egan, & Fan, 2016), cognitive control is a fundamental process that serves as a core component of broadly defined executive functions and higher level cognition such as intelligence (Chen et al., 2019). Cognitive control is involved in the efficient processing of conflict and is supported by an integrated cognitive control network (CCN), including regions such as the anterior insular cortex (AIC), anterior cingulate cortex (ACC), frontal eye field (FEF), and areas near and along the intraparietal sulcus (IPS) (T. Wu et al.,
2020). The AIC constrains the capacity of cognitive control (CCC), one of the important aspects of cognitive control (T. Wu et al., 2019). Convergent neuroimaging evidence has demonstrated the involvement of the AIC in cognitive control; however, the functional role of the AIC in cognitive control remains unknown.
Rationale for Study
The examination of cognitive control in individuals with lesions in any regions of the
CCN serves as a valuable approach to show a causal relation. Regretfully, lesion studies on the
AIC and cognitive control are sparse and show inconsistent results due to insensitive measurement to detect subtle deficits in cognitive control in neurological populations. Processing efficiency, an index that balances both performance accuracy and response time, can be used to assess cognitive control in patients with lesions in the AIC. In addition to conducting the human
1
lesion study, we also conducted a mouse study to examine the relationship between the AIC and cognitive control. The use of a model organism allowed invasive manipulation of the neuronal activity of the AIC in the mouse brain and included techniques of optogenetics and fiber photometry recording that provided high spatial and temporal resolution.
The goal of this study was to understand the functional role of the AIC in cognitive control. The current study contributes to a better understanding of the key structures underlying cognitive control and provides new insights into the neuromechanism of cognitive control. In addition, the current study serves as a starting point to examine the functional parcellation of the
CCN.
Research Questions and Hypothesis
The current study was designed to examine the functional role of the AIC in cognitive control based on human lesion-based evidence and a mouse model. The research questions of the present dissertation study examined 1) whether the AIC is necessary for cognitive control, and 2) how the AIC supports cognitive control, including whether the bilateral AIC is important for cognitive control and at what stage(s) (state uncertainty processing, execution of control, or motor generation) the AIC plays a critical role in cognitive control. The first research question was addressed by a human lesion study that included patients with focal lesion in the AIC. The second research question was addressed by a mouse study in which we used advanced techniques such as optogenetics and fiber photometry to manipulate and record the neuronal activity of the
AIC in mice. Based on empirical findings indicating that the AIC serves as a bottleneck of cognitive control (T. Wu et al., 2019), the research hypothesis of the current dissertation was that the AIC would play a causal role in efficient implementation of cognitive control. A comparative interference task was developed to measure the cognitive control of humans and mice. It was
2
predicted that 1) reduced performance and deficits in cognitive control would be observed in the patients with AIC lesions, 2) reduced performance and deficits in cognitive control would be observed in mice when the neural activity of the AIC was inhibited, and 3) signal changes would be detected when cognitive control was involved.
3
CHAPTER 2
Literature Review
Definition of cognitive control
For decades, the psychological construct, cognitive control, has been defined in multiple ways without a consensus (Mackie et al., 2013; see Nigg, 2017, for a review). Different terms have been used for related psychological constructs such as attention or attentional control, effortful control, executive control or executive control of attention, executive functions, inhibitory control, and self-control or self-regulation. Attention or attentional control refers to processes that select prioritized information for access to consciousness, which has been separated into three attentional functions: alerting, orienting, and executive control of attention
(Fan, McCandliss, Fossella, Flombaum, & Posner, 2005; Fan, McCandliss, Sommer, Raz, &
Posner, 2002; Mackie et al., 2013; Posner & Boies, 1971; Xuan et al., 2016). Effortful control, defined as the ability to inhibit dominant responses, monitor errors, and engage in planning, is a crucial construct in human development and an essential concept in psychopathologies (Rothbart
& Rueda, 2005). Executive control, or executive control of attention, minimizes distraction from perceptual levels, overcomes interference of thoughts, and suppresses prepotent responses (M.
Anderson & Green, 2001). Executive functions refer to a variety of top-down mental processes to accomplish goal-directed behaviors, including response inhibition, working memory, and cognitive flexibility or set shifting (see Diamond, 2013 for a review; Miyake et al., 2000).
Inhibitory control, interchangeably termed response inhibition, is one of the core components of executive functions and refers to intentional suppression of habitual responses for the purpose of goal-directed behaviors (see Diamond, 2013 for a review). A broader psychological concept of self-control or self-regulation is defined as the ability to voluntarily adjust cognitive and
4
behavioral responses to support the pursuit of long-term goals (Baumeister, Vohs, & Tice, 2007;
Duckworth, 2011; Fujita, 2011). All of these psychological constructs are interrelated and share a fundamental process based on their definitions and characteristics: coordination of thoughts and actions in support of accomplishment of goal-directed behaviors, defined as cognitive control in the current study. An information theory account of cognitive control (Fan, 2014) has quantified information processing in units of measurement called bits.
Cognitive control in human studies
Measurement of cognitive control. Cognitive control is typically measured by Stroop tasks (Stroop, 1935) and flanker tasks (Fan et al., 2002), which involve conflict processing by manipulation of congruent and incongruent conditions. For instance, in flanker tasks, individuals are required to make decisions about the direction of an arrow pointing either left or right. This arrow is flanked by two additional arrows on each side. The flankers, as the distractors, point in either the same direction (congruent condition) or the opposite direction (incongruent condition) of the target arrow. The difference in accuracy or response time (RT) between the congruent and incongruent conditions is called the conflict or interference effect. The conflict effect of RT is approximately 100 ms in normal populations.
Cognitive control is also measured by oddball tasks (Kiehl, Hare, Liddle, & McDonald,
1999), stop signal tasks (Logan, 1994; Logan & Cowan, 1984), and Go/No-go tasks (Menon,
Adleman, White, Glover, & Reiss, 2001; Nieuwenhuis, Yeung, van den Wildenberg, &
Ridderinkhof, 2003), in which one stimulus type occurs at a low probability (e.g., 20%) in series with another stimulus type that occurs at a high probability (e.g., 80%). For instance, in the typical oddball tasks, 80% of the stimuli as targets are intermixed with 20% of the stimuli as non-targets/oddballs. Targets and non-targets are both task relevant but are different in one of
5
their dimension features, such as color, shape, or size. Participants are requested to respond to the targets while ignoring the non-targets/oddballs. In another version of the oddball task, the oddball is manipulated by changing the probability of task-irrelevant features (Q. Wu et al.,
2015). The oddball effect is calculated as the difference between the standard and the oddball condition, which is approximately 30 ms in normal adults (Q. Wu et al., 2015).
Cognitive control network. A commonly activated network underlying cognitive control that is measured by a wide range of tasks is known as the cognitive control network (CCN), which consists of a frontoparietal network, a cingulo-opercular network, and subcortical structures (Fan et al., 2014; T. Wu et al., 2020; T. Wu et al., 2018). The frontoparietal network is composed of regions of the prefrontal and parietal cortices, such as the FEF and supplementary eye field, the mid frontal gyrus (MFG), areas near and along the IPS, and the superior parietal lobule (Corbetta, 1998; Fan et al., 2014). The cingulo-opercular network is also known as the saliency network in which the AIC and the ACC serve as key structures (Dosenbach et al., 2007;
Dosenbach et al., 2006). Subcortical structures subserving cognitive control include the thalamus and the basal ganglia (Fan et al., 2014; Koziol, 2014; Rossi, Pessoa, Desimone, & Ungerleider,
2009). The integrated role of the CCN and the functional connectivity of the CCN in cognitive control have been demonstrated in a growing number of functional magnetic resonance imaging
(fMRI) studies (Cole & Schneider, 2007); however, the functional parcellation of the CCN is not clear, including the distinct roles of regions in the large-scaled network.
Capacity of cognitive control. The amount of information that can be processed in a unit of time via cognitive control is not infinite (T. Wu et al., 2016), just as the active maintenance of items in working memory is limited (Fukuda, Awh, & Vogel, 2010). The capacity of cognitive control (CCC) is referred to as the maximum rate of information processing that is implemented
6
by cognitive control. The CCC is indexed by a perceptual decision-making task (backward masking majority function task, MFTM) (T. Wu et al., 2016). In the MFTM, cognitive load is manipulated by varying the bits of information that have to be processed, and exposure time is manipulated by changing the duration of the presentation of information (Chen et al., 2019; T.
Wu et al., 2016). The CCC has been quantified by the unit of bits per second (bps), estimating how many binary decisions can be made in a unit of 1 second (T. Wu et al., 2016). When cognitive load exceeds the CCC, the accuracy will start to drop. For each individual, the upper limit (i.e., the CCC) can be estimated based on the relationship between cognitive load and response accuracy using model fitting (T. Wu et al., 2016). The CCC develops as age increases and is approximately 1 to 5 bps in childhood and adolescence (Chen et al., 2020); it remains stable in adulthood and even in old age at approximately 3 to 5 bps (He et al., 2019; T. Wu et al.,
2016). Higher CCC has been associated with increased activation of the right AIC (T. Wu et al.,
2018; T. Wu et al., 2019), indicating that the AIC could be a key structure that constrains the
CCC. A subsequent lesion study showed that 1) lesions in the AIC, not in the ACC, led to reduction of the CCC; and 2) the simulated lesions of the AIC resulted in the reduction of global efficiency of the CCN, suggesting the specific role of the AIC as a bottleneck of cognitive control (T. Wu et al., 2019).
Processing efficiency of cognitive control. Processing efficiency may work as another key aspect of cognitive control, in addition to the CCC that represents the span of cognitive control. Cognitive control is typically reflected by the conflict effect measured in the flanker tasks (Fan et al., 2002; Trautwein, Singer, & Kanske, 2016; Xuan et al., 2016) and the Stroop tasks (Stroop, 1935). A greater conflict effect indicates poorer cognitive control. Typically, either
RT or accuracy is emphasized separately in conflict tasks, which may lead to a strategy of speed-
7
accuracy tradeoff: decisions are made slowly to obtain high accuracy or quickly regardless of low accuracy (see Heitz, 2014, for a review). Specifically, in lesion studies, in order to maintain high accuracy, patients with lesions displayed prolonged RT in both congruent and incongruent conditions, leading to no differences in the conflict effect compared to normal controls (Fellows
& Farah, 2005; Rinne et al., 2013). These findings indicate that using RT or accuracy separately may not detect the subtle deficit in cognitive control of lesion patients, and the dynamic relationship between response speed and accuracy should be reflected in the measurement of cognitive control. Processing efficiency, defined as performance accuracy within a unit of time, may be used as a sensitive measurement of cognitive control to investigate the involvement of the AIC in cognitive control. Processing efficiency balances the accomplishment of both high response speed and accuracy supported by cognitive control (Mackie et al., 2013). Like the CCC, processing efficiency of cognitive control, with the circumvention of speed-accuracy tradeoff, can be used to reflect the change of psychological processes underlying cognitive control in neurological patients.
Top-down and bottom-up cognitive control. The key structures in the CCN may have fine-grained functional distinctions in top-down and bottom-up cognitive control. Top-down cognitive control refers to a voluntary process that guides behaviors in light of internal goals
(Dalley, Everitt, & Robbins, 2011), whereas bottom-up cognitive control refers to an automatic process to influence behaviors that are driven by salient stimuli (Corbetta & Shulman, 2002). A dorsal frontoparietal network has been identified that supports goal-directed top-down cognitive control (Chiu & Yantis, 2009; Corbetta, Kincade, Ollinger, McAvoy, & Shulman, 2000), whereas a ventral frontoparietal network supports stimulus-driven bottom-up cognitive control
8
(Shomstein, 2012). Additionally, the temporoparietal junction is involved in the integration of top-down and bottom-up cognitive control (Q. Wu et al., 2015).
The insular cortex and its anterior part. The insular cortex of humans is a hidden lobe of the brain known as the island of Reil (Reil, 1809), which lies in the depth of the lateral sulcus
(see Binder, Schaller, & Clusmann, 2007; Flynn, 1999 for reviews). It is traditionally considered a paralimbic cortex (Mesulam & Mufson, 1982) or a limbic integration cortex (Augustine, 1996).
The central insular sulcus separates the insular cortex into an anterior part and a posterior part.
The insular cortex is divided into three subdivisions based on the cytoarchitectural features: the granular insular cortex with six classical layers, the dysgranular insular cortex with thinner layer
IV, and the agranular insular cortex without the external granular layer (II) and the internal granular layer (IV) (Mesulam & Mufson, 1985). The posterior part and the anterior part of the insular cortex are occupied by the granular subdivision and the agranular subdivision, respectively. The insular cortex is anatomically connected with the frontal (e.g., the ACC, the orbitofrontal cortex, and the medial prefrontal cortex), parietal, and temporal lobes (see Gogolla,
2017 for a review). The insular cortex receives and integrates inputs across modalities including auditory, somatosensory, olfactory, gustatory, visual, and interoceptive information (Avery et al.,
2015; Bamiou, Musiek, & Luxon, 2003; Heining et al., 2003; Mazzola, Isnard, & Mauguiere,
2005; Naghavi, Eriksson, Larsson, & Nyberg, 2007; X. Wang et al., 2019), and it is considered a hub that connects brain networks (Moran et al., 2013; T. Wu et al., 2019).
The anterior insular cortex (AIC), considered a limbic sensory region (see Craig, 2009 for a review), is implicated in various functions related to emotion and cognition. The AIC has been shown to be activated in a variety of processes (see Craig, 2009; Gogolla, 2017 for reviews) such as interoception (X. Wang et al., 2019; Zaki, Davis, & Ochsner, 2012); awareness of body
9
movement (Farrer & Frith, 2002; Tsakiris, Hesse, Boy, Haggard, & Fink, 2007); self-recognition
(Devue et al., 2007); vocalization and music (Peretz & Zatorre, 2005; Platel et al., 1997); emotional awareness (Feinstein et al., 2016; Gu, Hof, Friston, & Fan, 2013; W. K. Simmons et al., 2013); risk, uncertainty, and anticipation (Preuschoff, Quartz, & Bossaerts, 2008;
Sarinopoulos et al., 2010; Singer, Critchley, & Preuschoff, 2009); empathy (Gu et al., 2012;
Singer et al., 2004); visual and auditory awareness of the moment (Kosillo & Smith, 2010); time perception (Sterzer & Kleinschmidt, 2010); attention (Menon & Uddin, 2010; Nelson et al.,
2010); perceptual decision making (Chand & Dhamala, 2017); and cognitive control (Brass &
Haggard, 2007; Cole & Schneider, 2007; Dosenbach et al., 2007; Fan et al., 2014; Ramautar,
Slagter, Kok, & Ridderinkhof, 2006; T. Wu et al., 2018; T. Wu et al., 2019).
Structural and functional connectivity of the AIC. The structural connectivity of the
AIC has been extensively examined using diffusion tensor imaging in humans. The AIC has been shown to be connected with regions in the frontal lobe such as the superior, middle, inferior and orbital gyri; regions in the parietal lobe such as the superior parietal lobule, postcentral gyrus, and supramarginal gyrus; and regions in the temporal lobe such as the superior, middle, and inferior gyri (Cloutman, Binney, Drakesmith, Parker, & Lambon Ralph, 2012; Ghaziri et al.,
2017; Jakab, Molnár, Bogner, Béres, & Berényi, 2012). Structural connections have also been revealed between the AIC and the ACC (Ghaziri et al., 2017), which serve as key regions of the salience network and the CCN supporting the accomplishment of goal-directed behaviors. In addition, the AIC is connected with the caudate nucleus and subcortical regions such as the centromedian/parafascicular complex of the thalamus, which compromise thalamico-cortical loops for information processing (Eckert et al., 2012). The resting state studies have revealed the links of the AIC to the ACC (Taylor, Seminowicz, & Davis, 2009); to the superior, middle and
10
inferior frontal gyri (Cai, Ryali, Chen, Li, & Menon, 2014; Zhang, Ide, & Li, 2012); and to the middle and inferior temporal cortices (Cauda et al., 2011).
Lesions in the AIC and cognitive control. Neuroimaging evidence from fMRI studies has not demonstrate the causal relationship between the AIC and cognitive control, but lesion studies may provide insights into the causally necessary function of the AIC in cognitive control.
Although lesions in the AIC have been examined from the perspective of other functions such as the empathy for pain (Gu et al., 2012), sparse lesion studies have examined the role of the AIC in cognitive control or other related executive functions, and results have been inconsistent.
A neuropsychological study recruited 22 patients with low- or high-grade gliomas in the anterior insula and reported their pre-surgery symptoms related to spatio-temporal orientation, non-verbal intelligence, verbal/spatial short-term memory, language comprehension, noun and verb naming, phonological fluency and discrimination, word and pseudoword reading, repetition and writing, lexical decision, visuospatial planning and ability, and attention (Tomasino et al.,
2014). In this study, a majority of patients with tumors in the AIC presented cognitive deficits in pre-surgical neuropsychological testing, although no deficits were found in non-verbal intelligence. A follow-up examination after surgery showed that less than 20% of the patients with lesions in the right AIC reported cognitive changes in attention and confusion, while more than 60% of the patients with lesions in the left AIC had cognitive changes in phonologic paraphasia, speech arrest, and anomia. Although the evidence provided by this study was not empirical, the reported neuropsychological symptoms contributed to a basic knowledge of the cognitive performance in patients with tumors in the AIC.
A case study showed that a woman with lesion in the left AIC was impaired in fluency, cognitive flexibility, and conflict processing, but not in language, visuospatial perception, and
11
memory, which suggested that the left AIC is necessary for cognitive control (Markostamou,
Rudolf, Tsiptsios, & Kosmidis, 2015). The 45-year-old woman was tested by tasks measuring executive functioning, such as fluency, flexibility, attentional control, and inhibitory control; language, visuospatial perception, attention/working memory; and memory including visuospatial memory, verbal memory and learning, and logical memory. The raw score of Stroop tasks assessing attentional control and inhibitory control of this patient was less than the 10th percentile based on age- and education-matched Greek normative samples, indicating that deficits in conflict processing in which cognitive control was involved might have resulted from lesion in the left AIC.
In contrast, another study showed that perception, memory, shifting, and intellectual ability of patients with lesions in the right insular cortex were intact (Bar-On, 2003). In this study, three patients with lesions in the right insular cortex and 11 patients with lesions outside the neural circuitry of somatic state activation and decision making, the brain damage control group, were tested by tasks measuring cognitive intelligence, perception, memory, executive functioning, and personality/psychopathology. Executive functioning, related to cognitive control, was assessed by the Wisconsin Card Sorting Test, the Trail-making Test (TMT), and the
Controlled Oral Word Association Test. Specifically, the TMT was used to measure the ability of task-switching, in which the time (in seconds) to complete the task was reported as the test score, and a higher score indicated greater impairment. The executive functioning of patients with the right insular lesion was not significantly different from the functioning of the brain damage controls, indicating that the right insula may not serve as a key region supporting executive functioning.
12
In contrast, a study with a large sample of stroke patients (n = 144, mean age = 71 ± 15 years) showed that lesions within the left insular cortex were associated with deficits in the performance of a task-switching paradigm, denoting its critical role in flexible switching/shifting of attention (Varjacic et al., 2018). No normal controls or brain damage controls were included in the study. Specifically, 60 patients had lesions in the left insular cortex, 47 patients had lesions in the right insular cortex, and 37 patients had lesions in the bilateral insular cortex. A shape-based
TMT analogue was used for the assessment of executive dysfunction including two baseline tests and a set-switching test. The number of accurately connected shapes in the set-switching test was indexed as the performance of the specific executive functioning (i.e., shifting), and a composite score for executive functioning was computed by subtracting the accuracy of the set-switching test from the sum of the accuracy of the two baseline tests. Higher composite scores indicated poorer executive functioning. The number of accurately connected shapes in the set-switching test was negatively related to lesion size but was not related to age, indicating that lesions in the insula may impact the ability of shifting as a function of lesion size. Additionally, voxel-lesion- symptom mapping showed that the lesions in the left insula were related to higher composite scores, suggesting a critical role of the left insular cortex in flexible switching of attention. The advanced age of the participants and the absence of control groups may have limited the reliability of the results.
Another study showed that response inhibition was impaired in patients with damage in the left inferior frontal gyrus (IFG) and the insula compared to a control group of patients with lesions in the orbitofrontal cortex (Swick, Ashley, & Turken, 2008). In this study, response inhibition was measured using a Go/NoGo task in which task difficulty (easy or hard) was manipulated by varying the probability of NoGo trials (50% or 10%). Patients with lesions in the
13
left IFG and the insula had higher error rates in the easy and the hard conditions with even higher error rates in the hard condition compared to the control group, suggesting a critical role of the left IFG and the insula in inhibitory control. However, the relative contributions of the left IFG and the insula to inhibitory control were not separated in this study. The role of the insula in inhibitory control, which is a related function of cognitive control, remains unclear.
The discrepancy of findings in the lesion studies regarding the relationship between insula and cognitive control may be attributed to the following: 1) The lesion locations in the insula were not coherent (i.e., whether the lesions were focalized in the anterior part of the insula), 2) the laterality of the lesions in the insula was not consistent, 3) the sample sizes were variable, 4) the control groups were selected differently, and 5) the measurements of cognitive control were not sufficient for or sensitive to the detection of the deficits in lesioned patients. A large cohort of patients with focal lesions in the anterior part of the insula with balanced laterality (i.e., the AIC), together with both brain-damage controls and neurologically intact controls and a sensitive measurement of cognitive control, would provide an opportunity to examine the role of the AIC in cognitive control.
Cognitive control in mouse studies
The mouse as a model mechanism. The mouse has been widely used in genetic research to provide insights into complex diseases of humans such as diabetes and hypertension as well as disorders of cognitive functions. Given that humans and mice share a large amount of genetic, anatomical, and physiological features, mouse models are valuable in experimental studies
(Rosenthal & Brown, 2007). Benefits of using the mouse as a model mechanism also include cost effectiveness, high rate of reproduction, and short generation time and lifespan. Most importantly, recent techniques have been developed to manipulate the genome of the mouse
14
including transgenesis, single-gene knock-outs and knock-ins, conditional gene modifications, and chromosomal rearrangements (D. Simmons, 2008). Other scientific tools such as optogenetics and fiber photometry have been developed rapidly in the past decade, providing new approaches for examining neuronal circuits underlying a wide range of functions in millisecond scales.
Optogenetics combines genetic and optical methods to manipulate (activate or inhibit) the activity of specific populations of neurons with high-temporal and cellular precision in freely moving mammals (see Deisseroth, 2011; Fenno, Yizhar, & Deisseroth, 2011 for reviews). The optogenetic tools consist of channelrhodopsins (i.e., microbial opsins that conduct cations and depolarize neurons upon illumination for excitation) and halorhodopsins (i.e., microbial opsins that conduct chloride ions into the cytoplasm upon yellow light illumination for inhibition). Six steps are involved in the application of optogenetic techniques: 1) Select a genetic construct including a promoter to drive expression and a genetically encoding opsin (light-sensitive ion channel), 2) insert the genetic construct into viral vectors such as lentivirus and adeno-associated virus (AAV) vectors, 3) inject virus into region(s) of interest in the animal brain so that the opsins are expressed in a specific population of neurons based on the promoter, 4) implant the ferrule with fiber in the region(s), 5) turn on the laser or LED light of specific wavelength and power to open ion channels in neurons to activate or inhibit neuronal activities, and 6) record electrophysiological and behavioral results. Optogenetics provides a way to examine the causality between neural activity and functions of specific structures in a variety of domains such as sensory processing, motivation, and cognition (Aston-Jones & Deisseroth, 2013).
Calcium-based optical fiber photometry allows for signal recording of calcium fluorescence as a measurement of spiking activity of neurons and behavioral recording
15
simultaneously in freely moving animals, such as the method of electrophysiological recording
(Liang, Ma, Watson, & Zhang, 2017). The steps involved in fiber photometry are similar to the steps in optogenetics, but genetically encoded calcium indicators such as a newly developed
GCaMP6 are widely used to express in specific types of neurons (Kim, Jayaraman, Looger, &
Svoboda, 2014) instead of genetically encoding opsins. The GCaMP6 consists of ultra-sensitive protein calcium sensors with different response kinetics and sensitivity: 1) Fast GCaMP6 sensors
(GCaMP6f) have fast rise and decay times at the cost of some loss in sensitivity, 2) slow
GCaMP6 sensors (GCaMP6s) have higher sensitivity but slower rise and decay times, and 3) medium GCaMP6 sensors (GCaMP6m) have modest response kinetics and sensitivity (A. Chen et al., 2013). Although fiber photometry may not provide causal evidence when neural circuits that subserve different functions are examined, it provides new windows to understand relationships between neural signals and behaviors.
Homologies between humans and mice provide opportunities to examine cognitive control by using mouse models. The mouse models serve as an entry point to unravel neuronal underpinnings of structures such as the AIC in cognitive control (Gogolla, 2017). With advanced techniques, mouse models allow researchers to explore not only the functional implications of the structures but also the functional microcircuits and the functions of specific subpopulations of neurons (Gogolla, 2017).
Measurement of cognitive control. The five-choice serial reaction time task (5-CSRTT) is commonly used to measure cognitive functions including attention, inhibitory control, motivation, and rule-reversal learning in mouse studies (Papaleo, Erickson, Liu, Chen, &
Weinberger, 2012). In this task, mice are required to detect brief light flashes at one of five spatial locations to earn food rewards. Premature responses before the presentation of the target
16
stimulus are considered an index of impulsivity. Omission rate is computed as the percentage of omissions (responses made outside the response window) out of all trials and is referred to as an index of motivation. Accuracy of behavioral performance is the index of attention. In some variants of the 5-CSRTT, attentional load is manipulated by varying the size of the attentional field, interference level, and probability of stimuli presented at specific locations. The five- choice continuous performance test (5C-CPT), as a well-known variant of the 5-CSRTT, includes non-target trials in addition to the regular trials presented in the 5-CSRTT to measure response inhibition (Young, Light, Marston, Sharp, & Geyer, 2009). In the non-target trials, all lights in the five spatial locations are illuminated, but mice have to suppress the responses to them. The stop-signal task can also be used as a measurement of response inhibition/inhibitory control (Eagle & Robbins, 2003). In the stop-signal task, 80% of trials are Go trials while 20% of trials are Stop trials with a tone as a cue to inhibit prepotent motor responses. A cross-modal divided attention task has been designed to examine the brain network underlying sensory selection and distractor suppression (Wimmer et al., 2015). In this task, mice are informed of the trial modality by different cue sounds and are required to respond to the target stimulus in a single modality while ignoring the distractor stimulus in another modality. Divided attention is assessed by performance accuracy under conditions across different modalities. Although these tasks have been used to measure different cognitive functions that are related to a core component, that is, cognitive control, a simple but valid and reliable task that involves conflict processing should be developed to measure cognitive control in mice.
Neural substrates underlying cognitive control. The prefrontal cortex (PFC) and the
ACC are important neuroanatomical hubs underlying cognitive control in mammals including mice. The PFC is considered a fundamental structure that represents and produces mental and
17
internal goal-directed behaviors (Carlén, 2017). The frontal cortical regions have been shown to be directly connected to the visual, somatosensory, and auditory cortices by using both the anterograde and retrograde tracing approaches (Zhang et al., 2016). The dorsomedial prefrontal neurons encode reward-relevant information and project to different populations of neurons such as corticostriatal neurons and corticothalamic neurons to guide behaviors (Otis et al., 2017). In addition, thalamic reticular subnetworks receive biased control from the PFC to select sensory inputs for further processing (Wimmer et al., 2015). The ACC also plays an important role in cognitive control, especially top-down cognitive control in mice. Top-down rather than bottom- up inputs from the ACC are sent to the claustrum, and the signal changes are a function of cognitive load, indicating that top-down cognitive control is implemented by the ACC-claustrum circuitry (White et al., 2018). The ACC modulates sensory processing in the primary visual cortex via top-down cognitive control (Zhang et al., 2014). More specifically, the excitatory neurons in the dorsal ACC modulate top-down cognitive control to goal-directed behaviors
(Koike et al., 2016). These findings with regard to how the PFC and the ACC influence goal- directed behaviors by the modulation of cognitive control on other lower level sensory regions such as the claustrum and the thalamus may shed light on a potential network of cognitive control in mice similar to the CCN in humans.
Structural and functional features of the insular cortex. There are three subdivisions of the insular cortex of rodents: granular, dysgranular, and agranular, though the functions of these subdivisions are not fully understood (Butti & Hof, 2010; Maffei, Haley, & Fontanini,
2012). The granular and dysgranular insular cortex in mice contributes substantially to processing taste and visceral information, as well as representing anticipatory cues (Accolla &
Carleton, 2008; Katz, Simon, & Nicolelis, 2001; Kusumoto-Yoshida, Liu, Chen, Fontanini, &
18
Bonci, 2015; Oliveira-Maia et al., 2012; Samuelsen & Fontanini, 2017; Schiff et al., 2018; T.
Yamamoto, Yuyama, Kato, & Kawamura, 1985). The posterior part of the agranular insular cortex in mice, together with the granular and dysgranular insular cortex, is implicated in sensory integration (Gogolla, Takesian, Feng, Fagiolini, & Hensch, 2014). The agranular insular cortex is considered a homologous structure to the AIC in humans (Qadir et al., 2018). Both excitatory pyramidal neurons and GABAergic interneurons have been found in the agranular insular cortex
(K. Anderson et al., 2009; Gallay, Gallay, Jeanmonod, Rouiller, & Morel, 2011; Gogolla et al.,
2014; Ohara et al., 2003; K. Yamamoto, Koyanagi, Koshikawa, & Kobayashi, 2010). Although the insular cortex of rodents is analogous to humans, there is a marked difference between them.
A special cell type, that is, large and bipolar von Economo neurons, is presented in the insular cortex in human but not in rodents, which may contribute to empathy, social awareness, and self- control (see Allman et al., 2011 for a review).
The insular cortex is anatomically connected with a wide range of regions in the mouse brain (see Gogolla, 2017, for a review). The insular cortex is reciprocally connected with the sensory cortex, olfactory bulb, and thalamus, which are responsible for sensory processing and interoception. It also receives inputs from and sends outputs to the limbic system, such as the parahippocampus, hypothalamus, and amygdala, that supports emotional functions. The motivation, reward, and defensive systems (e.g., bed nucleus of the striaterminalis, nucleus accumbens, habenula, ventral tegmental area periaqueductal grey, and parabrachial nucleus) bear bidirectional connection with the insular cortex. The ACC, medial PFC, and orbitofrontal cortex as key structures in cognitive systems also have reciprocal links with the insular cortex. In addition, the insular cortex receives strong neuromodulatory inputs from the ventral tegmental
19
area (dopaminergic), the basal nucleus (cholinergic), the raphe nuclei (serotonergic), and the locus coeruleus (adrenergic).
Context for the proposed study
Accumulating evidence shows that the AIC is activated when cognitive control is involved, suggesting that the AIC plays a critical role in cognitive control from the neuroimaging perspective in human studies. However, the causal relationship between the AIC and cognitive control is elusive. By examining cognitive control in patients with lesions in the AIC, researchers can reveal the necessity of the AIC in cognitive control. In addition, a mouse model of cognitive control can deepen the understanding of the functional implications of the AIC. Although little direct evidence supports the role of the AIC in cognitive control in mice, the connectivity between the AIC and other regions that are involved in cognitive control may provide information on the functional role of the AIC in cognitive control in mice using newly developed techniques of optogenetics and fiber photometry.
20
CHAPTER 3
Anterior Insular Cortex is Necessary for Cognitive Control: A Human Lesion Study
Abstract
Efficient coordination of goal-directed mental operations requires the mechanisms of cognitive control. Growing neuroimaging evidence suggests that the anterior insular cortex (AIC) is one of the commonly activated regions underlying cognitive control and serves as a bottleneck of cognitive control. However, the causal relationship between cognitive control and the AIC has not been clearly demonstrated because of inconsistent findings that may result from insensitive measurement of cognitive control. In this lesion study, we examined processing efficiency as a sensitive measurement of cognitive control in patients with focal lesions in the AIC by using a visual post-target interference task. The patients with AIC lesions showed longer overall response time (RT), lower overall processing efficiency, and greater conflict effect of RT and processing efficiency. These findings indicate that lesions in the AIC lead to deficits in processing efficiency of cognitive control supporting a critical role of the AIC in cognitive control.
Keywords: anterior insular cortex, cognitive control, conflict processing, lesion, processing efficiency
21
Introduction
Cognitive control, the process that effectively and dynamically coordinates mental operations to guide goal-directed behaviors (Fan, 2014; Mackie et al., 2013), is supported by the cognitive control network (CCN), which consists of the anterior insular cortex (AIC), the anterior cingulate cortex (ACC), the frontal eye field (FEF), the areas near and along the intraparietal sulcus (IPS), and subcortical regions such as the thalamus (Fan et al., 2014; T. Wu et al., 2020; T. Wu et al., 2018). The CCN contributes substantially to high-level information processing as a core entity in the brain, with the key structures activating as a function of the demands for cognitive control (Fan, 2014; Fan et al., 2014; T. Wu et al., 2020; T. Wu et al.,
2018; T. Wu et al., 2019). Prior work has shown that the activation of the regions (e.g., AIC and
ACC) in the CCN is a function of cognitive loads (Fan et al., 2014) and that the activation of the
AIC is associated with the rate of information processing under time constraints (T. Wu et al.,
2019).
Although the activation of the AIC and the ACC is associated with cognitive control indicating their essential roles in information processing under cognitive control (Trautwein et al., 2016), the causal relationships between the AIC and the ACC and cognitive control are inconsistent. Cognitive control has been extensively studied by examining the conflict effect
(Chen et al., 2019; Fan et al., 2005; Fan et al., 2002; Mackie et al., 2013). Greater conflict effect indicates poorer cognitive control. Some studies have shown that lesions in the insular cortex are associated with deficits in the shifting of attention and response inhibition (Swick et al., 2008;
Varjacic et al., 2018), which are closely related functions of cognitive control. Furthermore, damage in the ACC has resulted in severe deficits in intention and spontaneous response production, and mild impairments in focused and sustained attention (Cohen, Kaplan, Moser,
22
Jenkins, & Wilkinson, 1999). But other evidence has shown that neither the AIC nor the ACC damages have resulted in a significant increase of conflict effects in comparison to normal controls (Fellows & Farah, 2005; Rinne et al., 2013; Vendrell et al., 1995). The negative results may be attributed to the insensitivity of indices that are used in detecting the deficits of cognitive control. In previous studies, conflict effects of RT or accuracy have been computed as the differences between congruent and incongruent conditions. A strategy of speed-accuracy tradeoff may be used in the conflict tasks so that in order to maintain high performance accuracy, the lesion patients display prolonged RT in both congruent and incongruent conditions, which leads to no changes in the conflict effects. To eliminate the influences of this strategy, an index that combines both RT and accuracy should be used as the measurement of cognitive control.
Processing efficiency, which examines the dynamical relationship between RT and accuracy, is not affected by the strategy of speed-accuracy tradeoff so that it can be used as an effective and sensitive behavioral index to reflect the nature of mechanism underlying cognitive control. Previous studies have shown that processing efficiency is an important index with high sensitivity in detecting impairment among clinical populations such as alcoholic groups (Lawton-
Craddock, Nixon, & Tivis, 2003; Nixon, Paul, & Phillips, 1998) and schizophrenia patients
(Schatz, 1998), and provides information to explain potential individual performance differences
(Rypma et al., 2006). According to processing efficiency theory, processing efficiency relies on the balance between performance effectiveness and the amount of effort or resources used to accomplish goal-directed behaviors (Derakshan & Eysenck, 2009). Lesions in the key structures of the CCN such as the AIC or the ACC may disrupt this balance by reducing performance effectiveness and/or increasing attention to task-irrelevant stimuli, which leads to less attentional resources to the concurrent task demands. Assessment of processing efficiency allows for a
23
scrutiny of cognitive control adjustment (i.e., modulation of the balance between performance effectiveness and attentional resources) in addition to separate examinations of RT and accuracy.
In the current study, it was hypothesized that the AIC, rather than the ACC, would be necessary for processing efficiency of cognitive control. To test this hypothesis, we measured and compared the processing efficiency of cognitive control in groups of patients with AIC lesions, patients with ACC lesions, patients with lesions in the temporal lobe as the brain-damage control, and neurologically intact individuals as the normal control using a visual post-target interference task. In light of the results of our previous study that the AIC lesion, rather than the
ACC lesion, was associated with deficits in the capacity of cognitive control (T. Wu et al., 2019), we predicted that the processing efficiency of cognitive control in patients with lesions in the
AIC (AIC group) would be lower than in neurologically intact individuals (NC group), and the conflict effect in the AIC group would be greater than in the NC group, but the ACC group would not be significantly different from the NC group.
Methods
Participants
Patients with focal unilateral AIC lesions (AIC group, n = 17) were included in the study.
Patients with focal unilateral ACC lesions (ACC group, n = 14) participated in the study to serve as an active control group because the ACC is also one of the important regions of the CCN.
Patients with focal unilateral temporal lobe lesions as the brain-damage controls (BDC group, n
= 10) were also recruited to justify that impaired cognitive control resulted from the AIC lesions rather than from the brain surgical procedures per se. Neurologically intact individuals as the normal controls (NC group, n = 42) also participated in the study to provide reference of normal performance. All AIC, ACC, and temporal lobe lesions resulted from surgical removal of low-
24
grade gliomas. All patients and normal controls were recruited from the Patient’s Registry of
Tiantan Hospital, Beijing, China, and local Beijing communities, respectively. The NC group was matched with the patient groups by age, education, and ethnicity (ps > 0.05). All participants had normal color vision and reported no history or current state of neurological or psychiatric conditions. All participants completed the visual post-target interference tasks to measure cognitive control, the Mini-Mental State Examination (MMSE; Cockrell & Folstein, 2002) to assess cognitive ability, and the Beck Depression Inventory (BDI; Schwab, Bialow, Clemmons,
Martin, & Holzer, 1967) to measure mood state. There was no significant difference in the scores of the MMSE and the BDI between each lesion group and the NC group (ps > 0.05). Two patients in the AIC group, three patients in the ACC group, and one patient in the BDC group were excluded from further analyses because they failed to complete the behavioral task due to severe clinical symptoms. The final sample consisted of 15 patients in the AIC group, 11 patients in the ACC group, 9 patients in the BDC group, and 42 adults in the NC group. The characteristics of all groups are listed in Table 1, including the sample size (n), lesion laterality, chronicity in months, age in years, gender, education in years, MMSE, and BDI for each group.
Only one participant in each group was left-handed, and all other participants were right-handed.
The study was approved by the Institutional Review Board of the Tiantan Hospital of the Capital
Medical University in Beijing. All research was performed in accordance with relevant guidelines and regulations. Written informed consent was obtained from each participant prior to participation.
25
Table 1. Participant Characteristics in the Human Lesion Study
Lesion Chronicity Age Gender Education Group n MMSE BDI laterality (months) (years) (years) NC 42 N/A N/A 34.60 21M/21F 13.57 28.74 1.79 (10.61) (2.12) (1.91) (3.04)
BDC 9 5L/4R 19.11 39.78 11.78 28.33 1.11 5M/3F (16.47) (14.00) (3.90) (2.35) (2.09)
AIC 15 11L/4R 11.73 36.20 13M/2F 13.47 29.60 3.60 (13.39) (7.70) (2.36) (0.91) (2.36)
ACC 11 4L/7R 28.27 39.82 7M/4F 12.36 29.27 3.73 (30.18) (5.40) (3.01) (1.19) (3.26)
Note. Standard deviation is presented in parentheses. NC = normal controls, BDC = brain- damage controls, AIC = anterior insular cortex, ACC = anterior cingulate cortex.
Lesion reconstruction
Brain regions with lesion of the AIC and ACC groups were identified and plotted individually onto an anatomical template of a normal control (ch2.nii, provided by MRIcron:
RRID: SCR_002403, http://www.cabiatl.com/mricro/mricro/index.html) by two neurosurgeons
(H. J. and R. Y.) who were blind to behavioral results. The group overlaps of multiple lesions were created for the AIC group (Figure 1a) and the ACC group (Figure 1b), respectively, using the MRIcron, with all lesions mapped on the right hemisphere.
26
a
z = -3 z = 0 z = 3 z = 6 z = 9 z = 12 b
z = 26 z = 28 z = 30 z = 32 z = 34 z = 36
0 100%
Figure 1. Lesion mapping for patients with unilateral lesions (a) in the anterior insular cortex (AIC group) and (b) in the anterior cingulate cortex (ACC group). Note. Colors indicate the percentage of overlap of lesions across patients. Post-target interference task
In the post-target interference task, all stimuli were presented on a dark gray background with a black fixation cross in the center throughout the task. A single trial started with a pre- target fixation (PTF) for 0-1000 ms, followed by a target square (0.3°×0.3°) presented for 50 ms on either the left or the right side (visual angle is 4°) of the center fixation cross. After 50 ms inter-stimulus-interval (ISI), a distractor square (0.3°×0.3°) was displayed for 50 ms either on the same location as the target square (congruent condition) or on the opposite location of the target square (incongruent condition), followed by a post-target fixation for (2850-PTF) ms.
Participants were requested to press a left or right button on a mouse to indicate the location of the target square and to ignore the distractor square as quickly and accurately as possible. The oddball ratio was manipulated by varying probabilities of two colors of squares (black or white)
27
that were irrelevant to the task (80% standard trials vs. 20% oddball trials). The colors of the target and distractor squares remained the same. Therefore, the experimental design was a 2
(congruency: congruent, incongruent) × 2 (oddball ratio: standard, oddball) with four conditions in total. The response window started from the onset of the presentation of the target square and ended at the offset of the post-target fixation. In total, each trial lasted 3000 ms. The PIT-V consisted of four blocks, with 120 trials in each block. Each block started and ended with a 3-s fixation period. The PIT-V lasted approximately 25 min. The schematic of the PIT-V is presented in Figure 2.
+ + or + +
80% standard 20% oddball 80% standard 20% oddball
Target + (50 ms) Distractor (50 ms) + Pre-ta rget Fix ation (PTF) + 0-1000 ms TDI = + 50 ms
Distractor + 2850 ms - PTF
+ +
50% congruent 50% incongruent
Figure 2. Schematic of the post-target interference task in the human study. After a variable pre-target fixation (PTF) of 0-1000 ms, a target square is presented on either the left or the right side of a central fixation for 50 ms. After a target-distractor-interval (TDI) of 50 ms, a distractor square is presented on either the same side (congruent condition) or the opposite side (incongruent condition) of the target for 50 ms, followed by a variable fixation period of (2850 ms – PTF). Standard squares are presented for both target and distractor in 80% of trials, while oddball squares are presented in 20% of trials.
28
Data analysis
Means and standard deviations of error rate and accuracy were calculated for each condition (congruent-standard, incongruent-standard, congruent-oddball, and incongruent- oddball). Trials with error response or with response time (RT) exceeding ± 2 SD of the mean
RT in each condition for each participant were removed from further analysis. In total, 8.40% of trials were excluded in the AIC group, 11.00% in the ACC group, 9.54% in the BDC group, and
6.73% in the NC group. For each participant, the efficiency of cognitive control was calculated as the ratio of accuracy over mean RT (in seconds) of the remaining trials for each condition.
Means and standard deviations of RT and efficiency for each condition were calculated. The oddball effect and interaction effect of RT, error rate, and efficiency were not significant for all groups; therefore, the conditions of standard and oddball were combined to present the results of congruent and incongruent conditions.
The conflict effect of RT, error rate, and efficiency was calculated by subtracting the performance of the congruent condition from that of the incongruent condition. The conflict effect of efficiency would be negative because theoretically the efficiency in the incongruent condition is lower than that in the congruent condition. To keep all differences between the congruent and incongruent conditions positive, we calculated change in efficiency instead of a negative conflict effect of efficiency. Typically, a greater conflict effect of RT or error rate, or a greater change in efficiency, suggests lower cognitive control ability. The oddball effect of error rate, RT, and efficiency was computed by subtracting the performance of the standard condition from the performance of the oddball condition. The interaction effect was computed by subtracting the oddball effect of the incongruent condition from the oddball effect of the congruent condition.
29
The non-parametric bootstrapping method (Hasson, Avidan, Deouell, Bentin, & Malach,
2003; Mooney & Duval, 1993) was used to assess the probability of observing a difference between two groups (AIC vs. NC group, ACC vs. NC group, BDC vs. NC group, AIC vs. BDC group, ACC vs. BDC group, and AIC group vs. ACC group) by chance because the sample sizes in the lesion groups were small, which did not meet the assumptions of parametric statistics. The bootstrapping procedure was conducted with 1,000 iterations for each index of performance
(e.g., comparing the conflict effect of RT between 15 patients with AIC lesions and 42 NC individuals). In each iteration, 1) a whole sample of all 57 participants combining both groups were created, 2) 42 participants were randomly selected from the whole sample as the surrogate
NC sample, 3) 15 participants were randomly selected from the whole sample as the surrogate
AIC group, and 4) the t-value (one-tailed, AIC < NC) of the difference between the two surrogate groups was calculated. After 1,000 iterations, the distribution of the t-values was obtained. The observed t-value of the performance difference between the original AIC and NC groups was calculated and compared along this t distribution. If the probability of obtaining the observed t-value along the permutated distribution of t-values was less than 5% (one-tailed), the difference between the AIC and NC groups was considered significant. Reported p values are one-tailed.
In addition, the Bayes factor (BF) was calculated for each comparison. A BF greater than
100 indicates decisive evidence for the alternative hypothesis (H1) that there is a real difference in the population, a BF greater than 3 suggests substantial evidence for the difference, and a BF less than 1/3 indicates substantial evidence for the null hypothesis H0 that there is no correlation in the population. A BF value ranging from 1/3 to 3 suggests insensitivity of the data to distinguish between the H0 and H1 (Wetzels, Raaijmakers, Jakab, & Wagenmakers, 2009).
30
Results
Comparisons between the AIC, NC, and BDC groups
The mean RT, error rate, and efficiency
The mean RT (upper panel), error rate (middle panel), and efficiency (lower panel) with standard error across conditions are shown in Figure 3a. The mean RT of the AIC group (598.83 ms; 95% CI [537.76, 653.75]) was significantly longer than that of the NC group (534.84 ms;
95% CI [506.89, 562.92]), p = 0.027. The mean error rate of the AIC group (0.040; 95% CI
[0.026, 0.057]) was marginally significantly greater than that of the NC group (0.027; 95% CI
[0.021, 0.034]), p = 0.065. The mean efficiency of the AIC group (1.67; 95% CI [1.49, 1.89]) was significantly lower than that of the NC group (1.87; 95% CI [1.77, 1.98]), p = 0.039. There were no significant differences in the mean RT (p = 0.120), error rate (p = 0.195), or efficiency
(p = 0.264) between the AIC group and the BDC group. The mean error rate of the BDC group was significantly greater than that of the NC group, p = 0.050, but there were no significant differences of the mean RT (p = 0.334) or efficiency (p = 0.160) between the BDC and NC groups.
Conflict effects
The mean RT (upper panel), error rate (middle panel), and efficiency (lower panel) with standard error for congruent and incongruent conditions are shown in Figure 3b. The conflict effect of RT (upper panel) and error rate (middle panel), and the change in efficiency (lower panel), are shown in Figure 3c. The conflict effect of RT of the AIC group (105.59 ms, 95% CI
[70.77, 147.92]) was significantly greater than that of the NC group (64.83 ms, 95% CI [53.42,
77.76]), p = 0.021. The conflict effect of error rate of the AIC group (0.046; 95% CI [0.028,
0.067]) was marginally significantly greater than that of the NC group (0.029; 95% CI [0.020,
31
0.040]), p = 0.078. The change in efficiency of the AIC group (0.35; 95% CI [0.27, 0.42]) was significantly greater than that of the NC group (0.28; 95% CI [0.24, 0.32]), p = 0.050. There were no significant differences in the conflict effect of RT (p = 0.315), error rate (p = 0.393), or the change in efficiency (p = 0.442) between the AIC and BDC groups. The conflict effect of efficiency of the BDC group was marginally significantly greater than that of the NC group, p =
0.079, but there were no significant differences in the conflict effect of RT (p = 0.127) or error rate (p = 0.210) between the BDC and NC groups.
ab c 800 800 160 * 700 * 700 120 600 600 80 500 500 RT(ms) RT(ms) 40
400 400 RT(ms) Conflict:
300 300 0 NC BDC AIC ACC C ICICICI NC BDC AIC ACC NC BDC AIC ACC 0.20 0.20 0.08
0.16 0.16 0.06 0.12 * 0.12 0.04 Error rate Error 0.08 rate Error 0.08 0.02 0.04 0.04 Conflict: Error rate Error Conflict:
0 0 0 NC BDC AIC ACC C I C ICICI NC BDC AIC ACC NC BDC AIC ACC 2.20 2.20 0.50 * * 2.00 2.00 0.40 1.80 1.80 0.30 1.60 1.60 Efficiency Efficiency 0.20 1.40 1.40 1.20 1.20 0.10 Change in efficiency 1.00 1.00 0 NC BDC AIC ACC CI CI CI CI NC BDC AIC ACC NC BDC AIC ACC
Figure 3. Behavioral performance of the NC, BDC, AIC, and ACC groups. (a) Mean RT (upper), mean error rate (middle), and mean efficiency (lower) across conditions are presented. (b) RT (upper), error rate (middle), and efficiency (lower) of congruent and incongruent conditions are presented. (c) Conflict effects of RT (upper) and error rate (middle), and change in efficiency (lower) are presented. Note. *p < 0.05; Error bars indicate standard error.
32
*p < 0.05
Oddball effects
The oddball effect of RT of the AIC group (9.40 ms, SD = 20.98) was not significantly different from that of the NC group (9.16 ms, SD = 11.91) or the BDC group (2.70 ms, SD =
15.70), ps > 0.05. There was no significant difference between the AIC group (0.005, SD =
0.022), the NC group (0.003, SD = 0.014), and the BDC group (0.000, SD = 0.022) regarding the error rate, ps > 0.05. No significant difference was found between the change in efficiency of the
AIC group (0.04, SD = 0.07), the NC group (0.03, SD = 0.05), and the BDC group (0.02, SD =
0.06), ps > 0.05.
Comparisons between the AIC, ACC, NC, and BDC groups
The mean efficiency of the ACC group (1.68; 95% CI [1.41, 1.91]) was marginally significantly lower than that of the NC group (1.87; 95% CI [1.77, 1.98]), p = 0.078, but there were no significant differences of the mean RT (p = 0.153) or error rate (p = 0.211) between the
ACC and NC groups. There were no significant differences of the mean RT (p = 0.273), error rate (p = 0.363), or efficiency (p = 0.288) between the ACC group and the BDC group. There were no significant differences of all the conflict or oddball effects between the ACC and NC groups, or between the ACC and BDC groups, ps > 0.05. The differences of all behavioral indices between the ACC and AIC groups were not significant, ps > 0.05.
Discussion
A necessary role of the AIC in the processing efficiency of cognitive control
Combined with considerable neuroimaging evidence showing that the activation of the
AIC is associated with cognitive control (Fan et al., 2014; T. Wu et al., 2019), the lesion-based findings in the current study provide causal evidence supporting a critical role of the AIC in cognitive control. In the current study, the greater conflict effect in the AIC group in comparison
33
to the NC group was mainly derived from the slower response time (AIC group vs. NC group:
632 ms vs. 551 ms) and lower processing efficiency (AIC group vs. NC group: 1.57 vs. 1.80) in the incongruent condition than in the congruent condition. According to the information theory account of cognitive control, the information to be processed or the uncertainty in the incongruent condition was 1 bit higher than in the congruent condition (Fan, 2014). The results indicated that the 1-bit uncertainty carried by the distractor in the incongruent condition could not be reduced or solved successfully and efficiently in the patients with lesions in the AIC due to deficits in the processing efficiency of cognitive control. Remarkably, the patients recruited in the current study had lesions in the AIC only in one hemisphere. But the impairment of cognitive control was still detected, suggesting that only one hemispheric AIC is not sufficient to support efficient processing via cognitive control even under relatively low uncertainty. More serious deficits in cognitive control, such as much higher error rate, much slower response time, much less processing efficiency, much greater conflict effects, or even incompletion of the behavioral task, may be observed in patients with bilateral lesions in the AIC.
The causality between the disruption of the AIC and the deficits in cognitive control may result from the dysfunction of the AIC in the modulation of the central executive network and the default mode network. Three distinct functional networks have been identified to support cognitive control: 1) a central executive network, consisting of dorsolateral prefrontal cortex and posterior parietal cortex; 2) the default mode network, consisting of ventromedial prefrontal cortex and posterior cingulate cortex; and 3) a salience network, consisting of ventrolateral prefrontal cortex, the AIC, and the ACC (Fox, Corbetta, Snyder, Vincent, & Raichle, 2006;
Greicius, Krasnow, Reiss, & Menon, 2003; Seeley et al., 2007). The central executive network and the salience network are commonly activated while the anticorrelated default mode network
34
is commonly deactivated when cognitive control is involved (A. Chen et al., 2013; Fan et al.,
2014; Power et al., 2011). The AIC, a key node in the salience network, has been demonstrated as a critical structure that drives or modulates the interaction between the central executive network and the default mode network (Goulden et al., 2014; Sridharan, Levitin, & Menon,
2008). Structural and functional changes in the AIC are associated with deficits in cognitive control and also lead to disruption in the modulation of the central executive network and the default mode network in patients with schizophrenia (Manoliu et al., 2013; Moran et al., 2013).
These findings may provide knowledge to understand the deficits of cognitive control in patients with lesions in the AIC in the current study. The impairment of cognitive control may be due to an inefficient AIC modulation of other large-scale networks, which can be supported by the findings in one of our previous studies that showed the association between simulated lesions in the AIC and decreased global efficiency of the CCN (T. Wu et al., 2019).
In the current study, the negative results regarding the difference between the AIC and
BDC groups may have resulted from small sample sizes in both groups (15 patients in the AIC group and 9 patients in the BDC group). If the sample size of the BDC group had been as large as the NC group, that is, nearly 3 times the sample size of the AIC group, a significant difference between the AIC and BDC group might have been detected using the bootstrapping procedure.
The temporal lobe is not a typical structure involved in cognitive control. Prior work has shown that lesions in the temporal lobe result in dysfunction of semantic word fluency (Martin, Loring,
Meador, & Lee, 1990). Therefore, patients with damage in the temporal lobe were included as the BDC group in our study. Insignificant results between the BDC and NC groups, and between the ACC and NC groups, showed that no deficits in cognitive control were found in either the
BDC or the ACC group, indicating the brain damage would not lead to impairment in cognitive
35
control and serving as indirect evidence to support the specific functional role of the AIC in cognitive control.
Distinctions between the roles of the AIC and the ACC
Considerable evidence has shown that the increase of activation in the ACC is associated with the increasing demand of cognitive control (Botvinick, Nystrom, Fissell, Carter, & Cohen,
1999; Fan et al., 2005; Fan et al., 2014; MacDonald, Cohen, Stenger, & Carter, 2000; T. Wu et al., 2018; Xuan et al., 2016); however, the necessary role of the ACC, a jointly activated region with the AIC, in cognitive control has not been consistently demonstrated similar to the inconsistent conclusions about the necessity of the AIC in cognitive control in prior work. A case study reported that a patient with bilateral ACC lesions showed specific deficits in attention during the subacute postoperative period, but performance of attention returned to normal spontaneously after several months, suggesting that the involvement of the ACC in cognitive control was important even if the role of the ACC in cognitive control would be compensated by other regions several months after surgery (Janer & Pardo, 1991). In another case study, the ability for response conflict detection in a patient with focal lesion in the rostral-to-mid dorsal
ACC was intact, but the ability to inhibit responses was impaired, indicating a specific role of the
ACC in response inhibition rather than in conflict monitoring. A previous study showed different deficit patterns in a patient with focal lesion in the right mid-caudal ACC and in another patient with focal lesion in the left rostral to mid-dorsal ACC (Swick & Jovanovic, 2002): damage in the right mid-caudal ACC resulted in no difference in the interference effect and overall performance, but the facilitation effect was decreased in comparison to normal controls. In contrast, damage in the left rostral to mid-dorsal led to deficits in set maintenance and response inhibition, indicating a potential functional role of the ACC in cognitive control. In addition,
36
severe deficits in intention and spontaneous response production, and mild impairments in focused and sustained attention, were found in 12 patients with focal lesions in the bilateral
ACC, indicating that response intention and focus attention were modulated by the ACC (Cohen et al., 1999). The ACC is not a structure supporting unitary function (Shidara & Richmond,
2002; Turken & Swick, 1999). The rostral ACC (rACC) has been shown to be necessary for reactive adjustments via cognitive control based on the findings that patients with damage in the rACC failed to dynamically regulate cognitive control (Di Pellegrino, Ciaramelli, & Làdavas,
2007).
These convergent findings provide suggestions of causality between the ACC and cognitive control; however, in other research, intact cognitive control was found in patients with lesions in the ACC. Lesions in the ACC were not associated with deficits in cognitive control measured by the Stroop task (Vendrell et al., 1995). In that study, 18 patients with lesions in the medial ACC (15 patients with unilateral lesion and three patients with bilateral lesion) were tested. No significant difference in the Stroop effects of RT or errors was detected between the patients and the controls, which may have resulted from the compensatory effects of the contralateral ACC or other related regions (Vendrell et al., 1995). The RT and error were examined separately in the Stroop task without considering the speed-accuracy tradeoff, which may be a potential factor for the absence of significant difference between the patients with lesions in the ACC and cognitive control. Another previous study examined the cognitive control of four patients with damage to the dorsal ACC (dACC) under speed-emphasized or accuracy- emphasized instructions (Fellows & Farah, 2005). However, no deficits in the patients with lesions in the dACC were observed in cognitive control measured by either the Stroop task or the
Go/No-go task for both conditions of instructions, indicating that the ACC was unnecessary for
37
cognitive control. Although the strategy of speed-accuracy tradeoff was considered, a more direct measurement that takes both RT and accuracy into account in a single index should be used to measure cognitive control for the examination of the functional roles of specific regions in the CCN.
Explanations for the discrepancy in findings across studies with regard to the role of the
ACC in cognitive control include 1) focalization of the lesions in the ACC was not consistent, 2) the sample sizes varied greatly, 3) the laterality of lesions was mixed, 4) different measurements were used to assess cognitive control, and 5) the postsurgical days before behavioral testing were not constant, which could also be used to interpret the inconsistency in the relationship between the AIC lesions and cognitive control in previous studies. The necessity of the ACC in cognitive control can be further tested by using processing efficiency as the measurement of cognitive control in a relatively large number of patients with laterality-balanced focal lesions in the ACC to advance the understanding of the CCN.
A functional dissociation in cognitive control between the ACC and the AIC has been examined by this systematic study that measures processing efficiency of cognitive control in both groups of patients with AIC lesions and ACC lesions. In the current study, no deficits of cognitive control were observed in the patients with lesions in the ACC, supporting the argument that the ACC and the AIC may be functionally dissociated in cognitive control in terms of processing efficiency. The AIC and the ACC may contribute to cognitive control by different mechanisms regardless of the co-activation of these two regions in previous neuroimaging studies. The AIC is considered a limbic sensory region (Craig, 2009; Critchley, Wiens,
Rotshtein, Ohman, & Dolan, 2004; Gu et al., 2012), whereas the ACC is a limbic motor region
(Craig, 2009). Extensive connections have been found between the AIC and the thalamus (Eckert
38
et al., 2012) and between the ACC and the motor cortex (see Paus, 2001, for a review). The AIC may serve as a key node that is responsible for efficient coordination of thoughts and actions via cognitive control, and the ACC may play an important role in the implementation of behaviors as a consequence of information processing via cognitive control.
Top-down and bottom-up cognitive control
Intact bottom-up cognitive control observed in the current study may have resulted from an insensitive measurement of oddball effect. In the post-target interference task, oddball was manipulated on a purely task-irrelevant feature, which could be readily ignored without effortful involvement of cognitive control. Therefore, the oddball effect of RT in the NC group was barely evident (~10 ms). In one of our previous studies, the manipulation of bottom-up cognitive control was also task irrelevant and was independent of top-down control (Q. Wu et al., 2015).
An oddball effect of 30 ms was observed for the high cognitive load condition in which the information to be processed was 2.58 bits, while no oddball effect (~ 0 ms) was observed for the low cognitive load condition in which the information to be processed was 1 bit, indicating that the oddball effect may only present when the cognitive source is occupied by other processes such as top-down cognitive control. In the current study, the information to be processed in the incongruent condition was 2 bits, which was lower than 2.58 bits in the high cognitive load condition. Therefore, the reason that a smaller oddball effect was observed in the current study may be due to the relatively low cognitive load in the incongruent condition, not to mention the even lower cognitive load in the congruent condition. Another behavioral task with higher cognitive load may be designed and used for further examination of the role of the AIC in bottom-up cognitive control or the interaction of top-down and bottom-up cognitive control.
39
No causal relationship was found in the AIC, and bottom-up cognitive control may be due to the fact that the AIC may serve as a high-level structure in the hierarchical architecture of cognitive control, which is responsible for top-down cognitive control but not bottom-up processing. In the previous study mentioned above, the increases in activation of the bilateral
AIC, bilateral FEF, and bilateral IPS were associated with recruitment of top-down cognitive control, whereas the increases in activation of the bilateral IPS and bilateral middle occipital gyrus were associated with the recruitment of bottom-up cognitive control (Q. Wu et al., 2015); this supports our argument that the AIC may play a role in top-down rather than bottom-up cognitive control. In the Q. Wu et al. (2015) study, the bottom-up process was manipulated by changing task-irrelevant features. It is worth noting that prior evidence also showed that the AIC was engaged in stimulus-driven orienting (Hahn, Ross, & Stein, 2006) and salience processing
(Seeley et al., 2007). The bottom-up cognitive control in these studies was measured by manipulating task-relevant features, indicating that the bottom-up cognitive control was not purely independent of top-down processing. Therefore, the AIC has been shown to be involved in bottom-up cognitive control in these studies. In future studies, tasks with pure manipulation of bottom-up cognitive control should be used.
Conclusion
The AIC, one of the key structures in the CCN, is necessary for cognitive control, especially for processing efficiency. The findings in the current study may have theoretical implications for the functional parcellation of regions in the CCN and practical implications for a wide range of clinical populations characterized by deficits in cognitive control.
40
CHAPTER 4
Anterior Insular Cortex is Critical for State Uncertainty Processing: A Mouse Study
Abstract
The neural substrates underlying cognitive control have been examined extensively in animal studies, including the prefrontal cortex, the anterior cingulate cortex, or subcortical regions such as the thalamic reticular nucleus. However, the role of the AIC, a key structure in the CCN of humans, in the cognitive control of mice remains unclear. In the current study, we aimed to examine whether both hemispheric AIC is important for cognitive control and at what stage the
AIC supports cognitive control. We first developed a paradigm (post-target interference task) with congruent and incongruent conditions for freely moving mice to measure cognitive control, which was indexed by the difference in accuracy between congruent and incongruent conditions.
Using the technique of optogenetics, the unilateral or bilateral AIC was inhibited after the cue sound (i.e., the start of trial) or during the presentation of both target and distractor stimuli. The calcium-based fluorescence change was recorded in the unilateral AIC of mice using a fiber photometry system when mice were performing the task. The accuracy of congruent conditions decreased when the AIC was unilaterally and bilaterally silenced before the cue sound and when the AIC was bilaterally inhibited during the presentation of target and distractor stimuli.
Significant calcium-based fluorescence changes were detected after the cue sound and when reward/correct response was made. These findings indicate that the AIC in both hemispheres is critical for cognitive control especially for state uncertainty processing.
Keywords: anterior insular cortex, cognitive control, optogenetics, fiber photometry, state uncertainty processing
41
Introduction
Cognitive control and its underlying neural mechanism have been extensively examined using mouse models. Cognitive control of mice is typically measured by the five-choice serial reaction time task (5-CSRTT; Papaleo et al., 2012), the stop signal task (Eagle & Robbins,
2003), and tasks with conflict processing (Wimmer et al., 2015). Using these tasks, as known homologies to human prefrontal cortex (PFC) and anterior cingulate cortex (ACC), the prelimbic cortex and cingulate cortex in mice have been shown to be engaged in controlling goal-directed behaviors by modulating sensory processing in the thalamus and the primary sensory cortex
(Koike et al., 2016; Wimmer et al., 2015; Zhang et al., 2014). Although the roles of the PFC and the ACC in mice have been well-established, how the anterior insular cortex (AIC) contributes to cognitive control of mice is still unknown. In human studies, the AIC serves as one of the key structures involved in cognitive control and ultimately influences behaviors. The insular cortex across rodents has been argued to share functional features with the human insular cortex, and therefore mouse models can provide insights on both the insular functions and the neural underpinning (see Gogolla, 2017, for a review). In mouse studies, the insular cortex plays an important role in a variety of processes implicated in regulatory, emotional, and cognitive functions as well as the integration of sensory, emotional, and cognitive contents (Gogolla et al.,
2014). The relationship between the anterior part of the insular cortex of mice and cognitive control needs to be specifically examined.
The structural connectivity of the AIC in mice provides knowledge about its potential role in cognitive control. The insular cortex receives cognitive signals from the ACC, medial
PFC, and orbitofrontal cortex, and also projects back to these critical regions of cognitive control
(see Gogolla, 2017 for a review), indicating that the AIC may also be involved in cognitive
42
control. In addition, rich interconnectivity has been shown between the AIC and the claustrum
(Qadir et al., 2018), a small and irregular structure hidden beneath the insula (Crick & Koch,
2005). The claustrum is implicated in the integration of salient information and the modulation of cognitive processes by broadcasting outputs across a wide range of higher level structures for cognition (Zingg, Dong, Tao, & Zhang, 2018) and supports the function of distractor suppressing
(Atlan et al., 2018). Information interchanges between the AIC and the claustrum suggest that the AIC may also play an important role in cognitive processing. Moreover, the claustrum receives and amplifies top-down signals from the ACC (White et al., 2018), indicating that like the ACC, the AIC may also regulate cognitive control by projecting top-down signals to lower level structures such as the claustrum.
In the current study, it was hypothesized that the AIC in mice would be critical in cognitive control as it is in humans. A compatible post-target interference task was designed for mice to include congruent and incongruent conditions by manipulating the locations of target and distractor stimuli. Optogenetic inhibition was exerted on either the unilateral or the bilateral AIC after the cue sound that initiated the trial or during the presentation of target and distractor in separate experiments. Additionally, fiber photometry recording was used to track the calcium- based fluorescence change in the post-target interference task. Based on the results of the human study, we predicted that the accuracy of congruent and incongruent conditions would decrease when the unilateral AIC was silenced and would drop more when the bilateral AIC was silenced for both inhibition periods because cognitive control was impaired when the AIC was inhibited.
Moreover, the fluorescence change during the presentation of the distractor stimulus in the incongruent condition would be markedly higher than in the congruent condition due to greater involvement of cognitive control.
43
Methods
Animals
Male and female mice (3 to 4 months old at the start of the experiments) were used for all experimental protocols. The C57BL/6 wild-type (WT) mice were used for the paradigm validation without any viral injection or ferrule implantation (n = 15). The WT mice (n = 13 in the experimental group and n = 9 in the control group) were used for the optogenetic experiments with viral or saline injection and fiber implantation. Transgenic mice (GCaMP6f×Rbp4-Cre) expressing Cre recombinase predominantly in isocortical layer 5 (pyramidal) excitatory neurons
(Rbp4-Cre) with genetically encoded calcium indicator (GCaMP6f) were used for fiber photometry experiments with ferrule implantation (n = 8). The GCaMP6f has a relatively fast rise time (~50 ms) and a fast decay time (~140 ms; T.-W. Chen et al., 2013; Kim et al., 2014).
An additional two GCaMP6f×Rbp4-Cre mice were used to examine the reward-related processing. All mice were maintained on a 12/12 light/dark cycle. Mice were trained overnight for the first three training sessions and then trained during the light cycle for 2 to 3 weeks. All behavioral tests were performed during the light cycle. Mice were group housed in cages containing two to four mice per cage during the training sessions (before surgeries), but were singly housed during the test sessions (after surgeries) to avoid damage of the fibers for each mouse. Virus expression and ferrule placement were verified at the end of the experiments. The ferrules implanted in three mice in the experimental groups dropped off before the data collection of experiments with bilateral inhibition. Two mice in the experimental group and one mouse in the control group in the optogenetic experiments, and one mouse in the fiber photometry recording, were excluded from further analysis due to ferrule misplacement. The final sample size consisted of 15 mice in the experiments of paradigm validation, 19 mice in the
44
optogenetic experiments (n = 11 in the experimental group and n = 8 in the control group) with unilateral inhibition, 16 mice in the optogenetic experiments (n = 8 in the experimental group and n = 8 in the control group) with bilateral inhibition, and seven and two mice in the experiments of fiber photometry recording for the examination of cognitive control and reward- related processing, respectively. All animal experiments were approved by the Queens College
Institutional Animal Care and Use Committee in accordance with National Institutes of Health guidelines for the responsible use of animals in research.
Chamber setup
Three sets of behavioral systems (Med Associates Inc., Vermont) were assembled in three customized black chambers (38.5 cm L × 25.6 cm W × 23.6 cm H) for behavioral training and testing. Each behavioral system consisted of a black chamber, a standalone USB interface, two bright yellow LEDs, two ultra-sensitive retractable levers with controllers, and two food dispensers. The two LEDs (0.79 cm each) were mounted horizontally (6.5 cm apart) on the middle of a wide wall of each chamber, with an outwards ultra-sensitive retractable lever and an outermost food hopper on each side. The diagram is shown in Figure 4a. Two food dispensers were placed outside the wide wall, connected to two corresponding food hoppers via a plastic tube to deliver the food (pellet) reward. All LEDs, levers, and food dispensers were linked to a standalone USB interface, communicating with the Med-PC-IV program running on a corresponding PC. There was an opening (diameter = 4 cm) on the top of each chamber cover for the optical patch cable to connect to the ferrule implanted in the brain of mice.
45
a LEDs Food hopper Food hopper
retractable lever retractable lever
Water bottle
b
Home-cage Home-cage Chamber 12 h 2 h Fasting period Training/Testing (water only) (food and water) (water only) sessions
Figure 4. (a) The diagram of chamber set-up. (b) The timeline of feeding periods. Training protocol
The daily timeline of training is shown in Figure 4b. Mice were placed into a fasting period of 12 hr with only water in the home cage before the behavioral training session each day.
After the training session, mice were put back into their home cage and subjected to 2-hr food restriction (water only), followed by a period of 10-hr ad libitum food and water. It was monitored daily that each mouse maintained at least 85% of their ad libitum body weight.
After 15-min habituation to the chamber, mice started the daily training session. The first five sessions were overnight training. A pellet was rewarded immediately when a correct response (lever press) was made. In the first two training sessions, mice had to learn the association between the illuminated LED and the corresponding lever to be rewarded. At the beginning of Training Session 1, the left LED was illuminated and the left lever was extended until the left lever was pressed 30 times. Then, the right LED was illuminated and the right lever was extended until the right lever was pressed 30 times. In Training Session 2, the two LEDs were illuminated alternatively with the corresponding lever extended. The LED was switched off and the lever was retracted for 50 ms after the lever was pressed. The inter-trial interval (ITI)
46
was 1 s. For the first two sessions, if both levers were not pressed for a total of 60 times over one night, the same session would take place on another night.
Starting from Training Session 3, a trial was initiated with a 50-ms cue sound. After 1000 ms, both levers were extended. After 500 ms, either the left or the right LED was illuminated as the target stimulus. Mice had to press the corresponding lever (i.e., the lever on the same side of the target stimulus) to be rewarded. The LED was switched off and both levers were retracted immediately with a correct response. A trial was terminated automatically when no response was made within 10 s and was recorded as an omitted trial. The ITI was increased to 20 s. To avoid side preferences developed by the mice, the target stimulus was presented at the same side as on the previous trial following an incorrect response or an omission. The parameters in the paradigm of Training Sessions 4 and 5 stayed the same as in Training Session 3 except that a trial was terminated when mice made no response within 5 s and 3 s, respectively. The number of trials with responses was calculated by subtracting the number of omitted trials from the number of total trials. Accuracy was computed as the ratio of the number of correct trials over the number of trials with responses. For these three sessions, each session was repeated until accuracy was no less than 60%.
Mice were trained to resolve conflict in Sessions 6 to 10. Each session lasted 5 hr in the daytime. In Session 6, the LED on one side was illuminated as the target stimulus for 3000 ms
(response window). After a target-distractor-interval (TDI) of 500 ms, the LED on the same side
(congruent condition) or the opposite side (incongruent condition) was illuminated for 3000 ms as the distractor stimulus. The response window was 6500 ms starting from the onset of the target stimulus to the offset of the distractor stimulus. Mice had to make a correct response for the target rather than for the distractor stimulus within the response window to be rewarded.
47
Trials with responses not made within the response window were recorded as omitted trials. The
ITIs were extended to 45 s in order to maintain the motivation of the mice and thus decrease the possibility of omitted trials. Omission rate was calculated as the ratio of omitted trials over the total trials. As long as the omission rate was no more than 30% and the accuracy was no less than
60% in the congruent condition, mice progressed to the next training session (Halassa et al.,
2014; Weller, Levin, Shiv, & Bechara, 2009; Wimmer et al., 2015). The duration of the target and distractor stimuli and the TDI were successively shortened in time in Training Sessions 7 to
10. In Training Session 10, the repetition of previous trials due to incorrect response or omission was removed, and all trials were randomized in terms of presentation side. The paradigm in
Training Session 10 was used as the behavioral testing task (Figure 5). The parameters of the paradigm for all training session are listed in Table 2. It took 1 to 4 days per session for the mice to reach the behavioral criteria. On average, 4 to 6 weeks were needed for the mice to learn the behavioral task.
Table 2. Parameters of the Paradigm in the Training Sessions for Mice
Training Session Target Target- Distractor Inter- Behavioral criteria session duration duration distractor duration trial (in (in interval (in interval hours) seconds) (TDI) seconds) (in (in seconds) seconds) 1 10 infinite NA NA NA (Sessions 1 and 2) At least 30 times of press 2 10 infinite NA NA 1 on either lever 3 10 10 NA NA 20 (Sessions 3, 4, and 5) 4 10 5 NA NA 20 Accuracy > 0.60 5 10 3 NA NA 20 6 5 3 0.5 3 45 (Sessions 6 to 10) 7 5 1 0.5 1 45 a. Omission rate < 0.3 8 5 0.5 0.2 0.5 45 b. Accuracy > 0.60 in the 9 5 0.3 0.1 0.3 45 congruent condition 10 5 0.2 0.1 0.3 45 Note. NA = not available.
48
Post-target interference task and paradigm validation
The schematic of the post-target interference task (PTI) is shown in Figure 5. A cue sound was presented by extending both levers for 50 ms and retracting the levers to initiate the trial. After an interval of 1000 ms, both levers were extended for 2000 ms (response window).
After the levers were extended for 500 ms, one LED (either on the left or the right) was illuminated as the target stimulus for 200 ms, followed by a TDI of 100 ms. One LED was illuminated as the distractor stimulus for 200 ms on the same location (congruent condition, 50% of trials) or on the opposite location (incongruent condition, 50% of trials) of the target stimulus.
Mice should respond to the location of the target stimulus (first illuminated LED) by pressing the corresponding lever within the response window while ignoring the distractor stimulus (second illuminated LED). When a correct response to the target stimulus was made within the response window, a pellet was rewarded immediately through the associated food dispenser to the food hopper. Otherwise, no food was given for either incorrect responses or omissions. Accuracy and response time (RT) were recorded by the Med-PC-IV program (Med Associates Inc., Vermont).
Each trial lasted 3550 ms, followed by an ITI of 45 s. There were two conditions (congruent vs. incongruent) in the task, with the location of illumination balanced. Trials of different conditions were presented in random order.
To validate the paradigm of PTI, mice were tested for three sessions. To obtain an adequate number of valid trials (at least 60 trials for each condition), the duration of the behavioral testing sessions was not fixed. Approximately 200 trials were included in each session.
49
Response window (2000 ms) Food is rewarded immediately after correct response
Distractor (200 ms)
Cue 1000 ms Lever 500 ms Target Lever (50 ms) extension (200 ms) 50% congruent retraction
100 ms 1500 ms ITI = 45 s ......
50% incongruent
Figure 5. Schematic of the post-target interference task in the mouse study. Note. TDI = target-distractor- interval; ITI = inter-trial-interval Virus
The virus AAV1-CKIIα-stGtACR2-FusionRed with titer ≥ 1×10¹³ vg/mL was a gift from
Ofer Yizhar (Addgene plasmid # 105669; http://n2t.net/addgene:105669;
RRID:Addgene_105669) (Mahn et al., 2018). This CaMKIIa-driven, soma-targeted guillardia theta anion-conducting channelrhodopsin fused to FusionRed has high efficiency in optogenetic silencing, which has higher photocurrents, much lower axonal excitation, high light sensitivity, and rapid kinetics (Mahn et al., 2018). The virus used in the current optogenetic experiments was diluted 1:10 in phosphate buffered solution (PBS) and was aliquoted and stored at -80 °C. The virus injection was followed immediately by optical ferrule implantation. At least 2 weeks were reserved for the virus to express prior to optogenetic testing sessions.
Surgery protocol
All surgical tools including blades, scissors, needles, clips, and tweezers were sterilized in an autoclave before surgeries. Mice were anesthetized with a mixture of ketamine (80 mg/kg) and xylazine (10 mg/kg) in the amount of 0.5% of body weight via intraperitoneal (IP) injection.
The scalp of each mouse was carefully shaved with an electric razor after no sign of reflexes.
50
Eye lubricant (Puralube™ ointment) was applied to whiskers of mice before shaving to avoid being cut negligently. Mice were placed into a stereotaxic frame (David Kopf Instruments) with head fixed by ear bars and were kept anesthetized through the nose using isoflurane (inducing at
5% before surgery and maintaining at 1.5% during surgery) during surgery. Mice were injected with bupivacaine in the amount of 1.50 mg/kg subcutaneously under the scalp to block nerve impulses. A heating pad was used to maintain body temperature at 32°C and eye lubricant
(Puralube™ ointment) was applied to both eyes to avoid irritation throughout the surgery. In the sterile surgical environment, the skull was exposed after the midline incision of skin on the top, and connective tissue was gently removed using a bent needle. Skin over the scalp was stretched to the sides for more exposure of the skull. Hydrogen peroxide was applied to the scalp by cotton swabs until the Bregma was clearly visible.
For the optogenetic experiments, a craniotomy (~0.5 mm diameter) was made above the injection sites. Virus (AAV1-CKIIα-stGtACR2-FusionRed) was injected at a rate of 46 nanoliter
(nl) per minute into the AIC bilaterally (AP +1.70 mm, ML +/– 2.60 mm, DV +3.75 mm) with a total volume of 1000 nl using a 33 G beveled needle (NanoFil syringe, World Precision
Instruments). The injection needle was not removed for 10 min in the brain to prevent a backflow of the virus to other areas. Following the viral injection, two ferrule optic fibers
(diameter = 200 µm) were implanted on the surface of targeted regions (~ 0.1 mm above the injection sites) and fixed with dental cement on the skull, together with a 2.5 mm ferrule stick and fiber cap to deliver light. After skin over the skull was sutured, mice were individually housed for at least 2 weeks before behavioral testing to allow for full recovery and viral expression. For the fiber photometry experiments, no virus was injected and two ferrule optic
51
fibers were implanted in the AIC bilaterally. All other steps for surgeries were the same as in the optogenetic experiments.
Optogenetics setup
The optogenetic system (Plexon Inc., Texas) used in the current study included compact
LED modules, dual LED commutators, and optical patch cables. The dual LED commutator was mounted 10 cm above the opening of the chamber cover, which could be equipped with magnetic compact LED module attachment bases. In this study, blue LED modules in the wavelength of
465 nm were used, emitting continuous, stable, and quick-responding light through the tip of a
200µm core. Each LED module was linked with a high durability optical patch cable that was connected to the ferrule implanted in the brain of mice to deliver blue LED light on the primary neurons affected by the virus (AAV1-CKIIα-stGtACR2-FusionRed) and thereby to silence the neuronal activity temporarily. A 4 Channel Optogenetic Controller with Radiant™, shared by all three sets of the PlexBright® optogenetic stimulation system, was connected to the Med
Associates standalone USB interface through a custom-built audio and video switch. The behavioral trial logic, including the timeline of illumination of two LEDs, on/off signals, and duration of optogenetic stimulation were controlled by the Med-PC-IV program (Med Associates
Inc., Vermont), while the strength and pattern of stimulation was controlled by software (Plexon
Inc., Texas) running on a host PC connected to the 4 Channel Optogenetic Controller. For optogenetic inhibition, the fiber cap on the scalp of each mouse was connected to the PC connector on the output port of the laser control box outside the chamber that was linked to a waveform generator.
52
Behavioral testing with optogenetic inhibition
Two or 3 days prior to the formal behavioral testing sessions, mice of both experimental and control groups were refreshed with the post-target interference task for 3 hr per day. In these pre-test sessions, mice were connected to optic cables to become acclimated. The ferrule on the patch cable was cleaned by alcohol before connecting with the ferrule on the mice. In the optogenetic experiments, the post-target interference task consisted of four conditions of trials: 2
(congruency: congruent, incongruent) × 2 (inhibition: non-inhibiting, inhibiting). On the inhibiting trials, blue light (465 nm, ~5.0 mW) was delivered to the targeted region from the laser via the optic fiber. All four conditions of trials were randomly displayed in the task.
The between-session manipulation comprised six conditions: 2 (inhibiting window: after- cue, target-and-distractor) × 3 (inhibiting laterality: left, right, bilateral). For each trial, blue light was delivered to silence the target site for 100 ms before the onset of the cue sound until the onset of the target stimulus (inhibiting duration = 1600 ms) as shown in Figure 6a or between the onset of the target stimulus and the offset of the distractor stimulus (inhibiting duration = 500 ms) as shown in Figure 6b. All mice were tested for two sessions with each inhibiting time window. In addition, the inhibiting laterality was manipulated. All mice were tested with both unilateral inhibition for four sessions (left- or right-hemispheric AIC for two sessions) and bilateral inhibition for two sessions. Each behavioral testing session lasted approximately 3 hr per day.
53
a Cue (50 ms) Lever extension Target (200 ms)
Distractor (200 ms) 1000Inh ibition: ms 1 600 ms 500 ms Lever retraction
TDI = 100 ms
Food is rewardedResponse immediately window after correct response
(2000 ms) b
Cue ITI = 45 s (50 ms) Lever extension Target (200 ms)
Distractor (200 ms) 1000 ms 500 ms Lever retraction Inhibition : 500 ms Food is rewardedResponse immediately window after correct response
(2000 ms)
ITI = 45 s
Figure 6. Optogenetic inhibition in the post-target interference task. (a) Inhibition window (duration = 1600 ms) after the cue sound. (b) Inhibition window (duration = 500 ms) during the presentation of target and distractor stimuli. Fiber photometry
The multi-channel fiber photometry system was the same as used in a previous study (Li et al., 2016). A dichroic mirror (MD498; Thorlabs) was used to reflect a laser beam from a 488- nm laser (OBIS 488LS; Coherent) to record the fluorescence signals. An optical fiber (230 mm outer diameter, 0.37 numerical aperture, 2 m long) was used to guide the light between the commutator and the implanted optical fiber. To avoid excessive bleaching, the laser power was adjusted 0.01-0.02 mW at the tip of optical fiber before each recording session. The GCaMP fluorescence was bandpass filtered (MF525-39, Thorlabs) and collected by a photomultiplier tube (R3896, Hamamatsu). The current output from the photomultiplier tube was converted to
54
voltage signals by an amplifier (C7319, Hamamatsu), which was further filtered through a low- pass filter (40 Hz cut-off; Brownlee 440). The analogue voltage signals were digitalized at 500
Hz and recorded by a Power 1401 digitizer and Spike2 software (CED, Cambridge, UK).
For fiber photometry recording, a behavior box (Thinker Tech Nanjing Biotech Co., Ltd,
Nanjing, China) was used to receive the transistor-transistor logic (TTL) signal from the Med-
PC-IV program (Med Associates Inc., Vermont) to the software of fiber photometry recording
(CamFiberPhotometry V2.0, Thinker Tech Nanjing Biotech Co., Ltd, Nanjing, China). At the beginning of each trial of the post-target interference task, a TTL signal was sent for the synchronization between the behavioral event (i.e., cue sound) and the fiber photometry recording. At the start of the first recording session, the fiber and the LED were pre-heated at
100% power for approximately 10 min to minimize photobleaching of the GCaMP proteins.
Photobleaching refers to a constant decrease in response from green fluorescence protein (GFP) which maintains at a low level (Sych, Chernysheva, Sumanovski, & Helmchen, 2019). The ferrule on the recording cable was cleaned by alcohol before connecting with the ferrule on the mice. Each recording session lasted at least 3 hr to obtain an adequate number of valid trials. The signal from the unilateral AIC was collected for six sessions with left- and right-hemispheric
AIC counterbalanced.
In the experiments of reward-related processing, 20% of the correct trials were not rewarded even if mice correctly responded to the target stimulus. For the remaining 80% correct trials, reward was delivered with a delay of 200 ms, 500 ms, or 800 ms when a correct response was made. The signal from the AIC in the right hemisphere was collected for three sessions for each mouse.
55
Histology
After the completion of all behavioral testing sessions, mice were anesthetized with 0.05- ml euthanasia solution and were perfused with 1´ PBS followed by ice-cold 4% paraformaldehyde. The brains were kept in 4% paraformaldehyde for ~6 hr and switched to 25% sucrose in 1´ PBS to keep at least 24 hr for dehydration. Brains were sliced into 50-µm coronal sections to verify the extent of virus expression and to examine whether the bilateral placement of fiber implanted in the AIC was accurate. Four to five brain slices were placed on a piece of precleaned microscope slides (Erie Scientific LLC, New Hampshire), and the slides were placed in 50%, 75%, 95%, 100%, and 100% ethyl alcohol, and xylene for 30 s, 30 s, 60 s, 60 s, 60s, and
60 s correspondingly for dehydration. Afterwards, four drops of toluene solution were placed on each slide and a piece of microscope cover glass was placed on top. A confocal microscope
(Fluoview FV10i, Olympus) was used to observe the virus expression and fiber placement on the next day after the slides became dry.
Data analysis
Paradigm validation. The overall omission rate, the omission rate for the congruent condition, and the omission rate for the incongruent condition were calculated. Trials with RT exceeding ± 2 SD of the mean RT in each condition (congruent, incongruent) were removed as outliers from further analysis. Mean accuracy and error rate in each condition was then calculated based on the remaining trials and was used to estimate cognitive control. The conflict effect was computed as the difference in error rate between the congruent and incongruent conditions (i.e., conflict effect = Error rateincongruent - Error ratecongruent). A greater conflict effect indicated lower cognitive control ability. Due to the small sample size in the study, nonparametric tests were used for all statistical analyses. Related-samples Wilcoxon signed rank
56
tests (Mann Whitney U tests) were used to examine the difference in omission rate, outlier rate, accuracy, and RT between the congruent and incongruent conditions. A one-sample Wilcoxon signed rank test was used to examine whether the accuracy of congruent and incongruent conditions was higher than chance level (i.e., 0.50) and whether the conflict effect of error rate was higher than 0. Kendall rank correlation coefficients of conflict effects between each pair of two sessions were computed for the calculation of reliability (3r/1+2r) where r was the average correlation coefficients of three pairs of two sessions.
Optogenetic inhibition. For each behavioral testing session, the overall omission rate was calculated as the probability of omission errors in the total trials. The overall outlier rate was calculated as the number of trials with RT exceeding ± 2 SD over the number of total trials. The outlier trials were excluded from further analysis. The error rate of each condition was computed as the proportion of trials with incorrect responses out of all remaining trials. The conflict effect was computed by subtracting the error rate in the congruent condition from the error rate in the incongruent condition. The inhibiting effect was computed by subtracting the error rate in the non-inhibiting condition from the error rate in the inhibiting condition. The overall omission rate, the overall outlier rate, the accuracy of each condition, the conflict effect, and the inhibiting effect were averaged across sessions for experiments with different inhibiting windows, for experiments with unilateral inhibition, for experiments with bilateral inhibition, and for different groups of mice. Due to non-normal data distribution and small sample sizes, non-parametric tests were used for statistical analysis. Specifically, a one-sample Wilcoxon signed rank test was used to examine whether the accuracy of congruent and incongruent conditions was higher than chance level (i.e., 0.50) and whether the conflict effect of error rate was higher than 0 without
57
any inhibition on the AIC, and related-samples Wilcoxon signed rank tests were used for single comparison.
Fiber photometry recording. The raw data were first smoothed with a low-pass
Butterworth filter (2 Hz) with a default 6th order, followed by segmentation based on behavioral events within individual trials such as the onset of cue sound, the extension of levers, the presentation of target stimulus, the presentation of distractor stimulus, response, and reward delivery in correct trials (Figure 7). Trials with RTs exceeding ± 2 SD were excluded from further analysis. The remaining trials were separated into four different types: correct-congruent, correct-incongruent, incorrect-congruent, and incorrect-incongruent. The baseline fluorescence signal (F0) was averaged over 1.5 s starting from 0.5 s preceding the trigger events (i.e., onset of cue sound). The fluorescence change rate (FCR) was calculated as (F - F0)/F0. The comparison between the FCR for specific behavioral events followed the method in a previous study (Li et al., 2016): 1) The 95% fraction of the baseline FCR was defined as the up-threshold value and the 5% fraction as the low-threshold value within each event; 2) the local signal peaks above the up-threshold value were averaged as the activation amplitude of the FCR for a certain event, and the time points above the up-threshold were summed as the activation duration; 3) the local signal peaks below the low-threshold value were averaged as the inhibition amplitude of the FCR for a certain event, and the time points below the low-threshold were summed as the inhibition duration; 4) the response amplitudes of the FCR for the presentation of target stimulus and distractor stimulus were compared between the congruent and incongruent conditions for correct and incorrect trials by the non-parametric related-samples Wilcoxon signed rank tests, respectively; 5) the average activation amplitude, activation duration, inhibition amplitude, and inhibition duration were averaged across the behavioral sessions for different behavioral events
58
of each mouse, and then were averaged across mice; and 6) the activation amplitude, activation duration, inhibition amplitude, and inhibition duration of different behavioral events for correct and incorrect trials were compared with 0 using the one-sample Wilcoxon signed rank test.
cue sound levers out end of trial (-1.55 to 1.5 s) (at -0.5 s) (at 2 s)
baseline signal F0 target distractor (-3.55 s to -2.05 s) (0 to 0.2 s)(0.3 to 0.5 s)
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 ... 8
Time (seconds)
Figure 7. Timeline for the analysis in the fiber photometry recording. The baseline fluorescence signal (F0) was averaged over 1.5 s starting from 0.5 s preceding the trigger events (i.e., onset of cue sound). The fluorescence change rate was calculated as (F - F0)/F0. For the experiments of reward process, correct trials were separated into two types: correct-no-reward and correct-reward, followed by a response-locked analysis. For the correct- reward trials, data were first analyzed with delivery locked. These trials were further separated by delayed intervals (i.e., 200 ms, 500 ms, and 800 ms). Within each of these trials, signals were segmented based on behavioral events of response making and food delivery.
Results
Paradigm validation
The descriptive statistics are presented in Table 3. The overall omission rate was 21.57%.
There was no significant difference in the omission rates between the congruent condition
(20.90%) and the incongruent condition (22.21%), p = 0.232. On average, 151 trials were responded to in the testing session, and 5.95% of trials were excluded as RT outliers. There was no significant difference in the outlier rates between the congruent condition (6.12%) and the incongruent condition (5.81%), p = 0.184. The accuracy of both the congruent condition (0.61)
59
and the incongruent condition (0.55) was higher than the chance level of 0.50, ps = 0.001. The accuracy of the congruent condition was significantly higher than the accuracy of the incongruent condition, p = 0.001. The conflict effect of error rate (0.06) was significantly higher than 0, p = 0.001. There was no significant different in RT between the congruent condition (340 ms) and the incongruent condition (350 ms), p = 0.550. The reliability of conflict effects across sessions of the paradigm was 0.52.
Table 3. Mean and Standard Deviation (SD) of the Number of Total Trials, and Omission
Rate, Outlier Rate, Accuracy, and RT for All Conditions (Overall), Congruent Condition
(Cong), and Incongruent Condition (Incong)
Outlier Total Omission rate Accuracy RT rate trials overall cong incong overall cong incong cong incong Mean 199.02 0.22 0.21 0.22 0.06 0.61 0.55 0.34 0.35 SD 25.01 0.08 0.08 0.09 0.01 0.03 0.03 0.11 0.10
Optogenetic inhibition
Viral expression and the sites of ferrule placement for the experimental group are presented in Figure 8. The overall omission rates and the overall outlier rates for different experiments and groups are presented in Table 4. There was no significant difference in the overall omission rate and outlier rate between the experimental group and the control group in the experiments of unilateral or bilateral inhibition, ps > 0.05. The accuracy of each condition for different experiments and groups are also presented in Table 4.
60
AIC
Bregma +1.94 mm
Bregma +1.70 mm
Bregma +1.54 mm
Figure 8. Viral expression and sites of ferrule placement in all experiments. Left top corner: viral expression in the AIC. Ferrule placement is presented in the diagram of brain slice at Bregma AP +1.94 mm, +1.70 mm, and +1.54 mm. Dots in red represent bilateral viral expression (n = 8) in the AIC, and dots in yellow represent an additional three mice with only unilateral viral expression. Dots in blue represent the control group (n = 8). Dots in green represent the group in the fiber photometry recording (n = 7). Table 4. Mean (SD) of the Overall Omission Rate and the Overall Outlier Rate for
Different Experiments and Groups
Experimental group Control group Inhibiting windows Laterality Omission Outlier Omission Outlier rate rate rate rate Unilateral 0.22 (0.07) 0.28 (0.10) 0.18 (0.04) 0.23 (0.06) After-cue-sound Bilateral 0.20 (0.07) 0.24 (0.09) 0.20 (0.05) 0.26 (0.09) Target-and- Unilateral 0.22 (0.05) 0.27 (0.06) 0.17 (0.05) 0.21 (0.06) distractor Bilateral 0.17 (0.08) 0.22 (0.09) 0.14 (0.03) 0.18 (0.09)
61
Results of the experimental group
Unilateral inhibition after the cue sound (Figure 9a) and during the presentation of target and distractor (Figure 9b). The accuracy of the inhibiting condition was significantly lower than the accuracy of the non-inhibiting condition in the congruent trials when the AIC was unilaterally inhibited after the cue sound, p = 0.046. For both the non-inhibition and inhibition conditions, the accuracy of the congruent conditions was higher than that of the incongruent conditions, ps < 0.05.
abExperimental group: Experimental group: 0.80 unilateral inhibition 0.80 unilateral inhibition after cue sound during target and distractor
0.75 0.75 ** ** * 0.70 0.70 * * 0.65 0.65
0.60 0.60
0.55 0.55 Accuracy 0.50 0.50 Accuracy
0.45 0.45
0.40 0.40
0.35 0.35
0.30 0.30 No Yes No Yes No Yes No Yes congruent incongruent congruent incongruent
Figure 9. Accuracy of conditions in the experimental group with/without unilateral inhibition on the AIC. (a) after the cue sound and (b) during presentation of target and distractor stimuli. Note. No = without inhibition; Yes = with inhibition. Error bars indicate standard error. **p < 0.005. *p < 0.05. Bilateral inhibition after the cue sound (Figure 10a) and during presentation of target and distractor (Figure 10b). The accuracy of the inhibiting condition was significantly lower than that of the non-inhibiting condition in the congruent trials when the AIC was bilaterally inhibited after the cue sound (p = 0.012) and during presentation of target and distractor (p = 0.011), but there were no inhibition effects in the incongruent trials for both
62
inhibiting periods, ps > 0.05. Only in the non-inhibition conditions, the accuracy of congruent conditions was higher than that of incongruent conditions for both inhibiting periods, ps < 0.05. abExperimental group: Experimental group: 0.80 bilateral inhibition 0.80 bilateral inhibition after cue sound during target and distractor 0.75 0.75 * * * * 0.70 0.70
0.65 0.65
0.60 0.60
0.55 0.55 Accuracy Accuracy 0.50 0.50
0.45 0.45
0.40 0.40
0.35 0.35
0.30 No Yes No Yes No Yes No Yes congruent incongruent congruent incongruent
Figure 10. Accuracy of conditions in the experimental group with/without bilateral inhibition on the AIC. (a) after the cue sound and (b) during the presentation of target and distractor stimuli. Note. No = without inhibition; Yes = with inhibition. Error bars indicate standard error. *p < 0.05. Results of the control group Unilateral inhibition after the cue sound (Figure 11a) and during presentation of target and distractor (Figure 11b). There were no inhibition effects for either the congruent or the incongruent conditions when the AIC was unilaterally inhibited during either inhibiting period, ps > 0.05. For both non-inhibition and inhibition conditions, the accuracy of the congruent conditions was higher than that of the incongruent conditions, ps < 0.05.
63
abControl group: Control group: 0.80 unilateral inhibition 0.80 unilateral inhibition after cue sound during target and distractor
0.75 0.75
0.70 * 0.70 * * 0.65 * 0.65
0.60 0.60
0.55 0.55 Accuracy Accuracy 0.50 0.50
0.45 0.45
0.40 0.40
0.35 0.35
0.30 0.30 No Yes No Yes No Yes No Yes congruent incongruent congruent incongruent
Figure 11. Accuracy of conditions in the control group with/without unilateral inhibition on the AIC. (a) after the cue sound and (b) during presentation of target and distractor stimuli. Note. No = without inhibition; Yes = with inhibition. Error bars indicate standard error. *p < 0.05. Bilateral inhibition after the cue sound (Figure 12a) and during presentation of target and distractor (Figure 12b). Similar to the results of unilateral inhibition in the control groups, there were no inhibition effects for either the congruent or the incongruent conditions when the AIC was unilaterally inhibited during either inhibiting period, ps > 0.05. For both non- inhibition and inhibition conditions, the accuracy of congruent conditions was higher than that of incongruent conditions, ps < 0.05.
64
abControl group: Control group: bilateral inhibition bilateral inhibition after cue sound during target and distractor
0.80 0.80 * * 0.75 0.75
* 0.70 0.70 *
0.65 0.65
0.60 0.60
0.55 0.55 Accuracy Accuracy 0.50 0.50
0.45 0.45
0.40 0.40
0.35 0.35
0.30 0.30 No Yes No Yes No Yes No Yes congruent incongruent congruent incongruent
Figure 12. Accuracy of conditions in the control group with/without bilateral inhibition on the AIC. (a) after the cue sound and (b) during presentation of target and distractor stimuli. Note. No = without inhibition; Yes = with inhibition. Error bars indicate standard error. *p < 0.05. Fiber photometry
Target-locked results. The FCR (i.e., ∆F/F0) of correct trials (black line) and incorrect trials (red line) is presented in Figure 13. The yellow line represents the onset of the cue sound, and the magenta line represents the onset of lever extension. The two grey bars indicate the presentation of the target stimulus and distractor stimulus, respectively. Grey shaded areas indicate the SEM. The results of activation amplitude, activation duration, inhibition amplitude,
65
or inhibition duration for correct and incorrect trials are presented in Table 5. For the correct trials, 1) all these indices were significantly higher than 0 in the cue-to-lever period and the lever-to-target period, ps < 0.05, indicating significant signal changes of the AIC in these two periods; 2) in the periods of target presentation and distractor presentation, only inhibition amplitude and duration were significantly different from 0, ps < 0.05, indicating decreased signal changes in the AIC in these two periods. Similarly, for the incorrect trials, significant signal increases of the AIC were found in the cue-to-lever period and the lever-to-target period (ps <
0.05), and significant signal decreases were found in the periods of target presentation and distractor presentation (ps < 0.05).
target-locked
20
10 (%) 0 0 F ∆ F/ -10
-20
-4 -2 02468 Time (seconds)
Figure 13. Target-locked averaged calcium transient in response to events in correct trials (black line) and incorrect trials (red line) across time. The yellow and magenta vertical lines represent the onsets of the cue sound and the lever extension, respectively. Grey bars represent the presentation of target and distractor with a 100-ms interval in between. Grey shaded areas indicate the SEM. The FCR of the congruent condition (solid line) and the incongruent condition (dashed line) is presented for the correct trials (in black lines) in Figure 14a and for the incorrect trials
(in red lines) in Figure 14b. There were no significant differences in these indices for all the event periods (cue-to-target, lever-to-distractor, target presentation, or distractor presentation)
66
between the congruent and the incongruent conditions for either correct or incorrect trials, ps >
0.05.
Table 5. The Mean and SD of Activation Amplitude, Activation Duration, Inhibition
Amplitude, and Inhibition Duration in the Event Windows of Cue-To-Lever, Lever-To-
Target, Target Presentation, and Distractor Presentation for the Correct and Incorrect
Trials
Activation Activation Inhibition Inhibition amplitude duration amplitude duration (s) (s) 0.05 0.49 -0.04 0.41 cue-to-lever (0.03) (0.24) (0.03) (0.20) 0.04 0.24 -0.04 0.22 lever-to-target (0.02) (0.19) (0.02) (0.18) correct trials 0.001 0.01 -0.08 0.20 target presentation (0.004) (0.04) (0.04) (0.00) 0.001 0.03 -0.08 0.20 distractor presentation (0.004) (0.07) (0.06) (0.00)
0.04 0.54 -0.03 0.32 cue-to-lever (0.02) (0.24) (0.03) (0.23) 0.03 0.24 -0.03 0.18 lever-to-target (0.03) (0.20) (0.03) (0.14) incorrect trials 0.001 0.01 -0.07 0.16 target presentation (0.003) (0.03) (0.06) (0.07) 0.001 0.03 -0.07 0.16 distractor presentation (0.003) (0.07) (0.06) (0.08)
67
a target-locked
20
10 (%) 0 0 F ∆ F/ -10
-20
-4 -2 02468 Time (seconds) b target-locked
20
10 (%) 0 0 F ∆ F/ -10
-20
-4 -2 02468 Time (seconds)
Figure 14. Target-locked averaged calcium transient in response to events in the congruent condition (solid line) and the incongruent condition (dashed line) in (a) correct trials (black) and (b) incorrect trials (red) across time. The yellow and magenta vertical lines represent the onsets of the cue sound and the lever extension, respectively. Grey bars represent the presentation of target and distractor with a 100-ms interval in between. Grey shaded areas indicate the SEM.
68
Response-locked results. The FCR (i.e., ∆F/F0) of correct trials (black line) and incorrect trials (red line) is presented in Figure 15 where the cyan line represents the time when a response was made and grey shades indicate the SEM. The results of activation amplitude, activation duration, inhibition amplitude, or inhibition duration for the periods of 0 to 4 s and 4 to 8 s after a response are presented in Table 6. Significant signal increases of the AIC in the period of 0 to 4 s after a response were detected in the correct trials only, p = 0.018. The FCR of the congruent condition (solid line) and the incongruent condition (dashed line) is presented for the correct trials (black lines) in Figure 16a and for the incorrect trials (red lines) in Figure 16b.
There were no significant differences in these indices for either period after a response between the congruent and the incongruent conditions for either correct or incorrect trials, ps > 0.05.
response-locked
20
10 (%) 0 0 F ∆ F/ -10
-20
-4 -2 02468 Time (seconds)
Figure 15. Response-locked averaged calcium transient in correct trials (black line) and incorrect trials (red line) across time. The cyan vertical line represents the time point when response was made. Grey shaded areas indicate the SEM.
69
Table 6. The Mean and SD of Activation Amplitude, Activation Duration, Inhibition
Amplitude, and Inhibition Duration in the Event Windows of 0 to 4 s and 4 to 8 s after
Response for Correct and Incorrect Trials
Activation Activation Inhibition Inhibition
amplitude duration (s) amplitude duration (s) 0.13 2.23 -0.06 1.54 0 to 4s (0.09) (1.08) (0.05) (0.99) correct trials 0.02 0.92 -0.10 2.97 4 to 8s (0.03) (1.47) (0.09) (1.45)
0.01 0.69 -0.07 3.10 0 to 4s (0.02) (0.99) (0.05) (1.05) incorrect trials 0.04 2.53 -0.04 1.28 4 to 8s (0.04) (1.92) (0.05) (1.89) a response-locked
20
10 (%) 0 0 F ∆ F/ -10
-20
-4 -2 02468 Time (seconds) b response-locked
20
10 (%) 0 0 F ∆ F/ -10
-20
-4 -2 02468 Time (seconds)
Figure 16. Response-locked averaged calcium transient in response to events in the congruent condition (solid line) and the incongruent condition (dashed line) in (a) correct trials (black) and (b) incorrect trials (red) across time. The cyan vertical line represents the time point when response was made. Grey shaded areas indicate the SEM.
70
Reward-related processing. The FCR (i.e., ∆F/F0) of correct-reward with delayed reward interval of 200 ms (blue line), 500 ms (yellow line), and 800 ms (magenta line), correct- no-reward (green line), and incorrect trials (red line) is described in detail in the main text and is also presented in Figure 17 where the cyan line represents the time when a response was made, the blue, yellow, and magenta vertical lines represent food delivery for each delayed reward respectively, and grey shades indicate the SEM. In specific, for the correct-reward trials, the signals started to increase around 800 ms and to decrease around 1.3 s after the food was delivered, which were consistent with delivery-locked results across trials with different delayed reward intervals (see Figure 18 where the purple line represents food delivery). The up-and- down pattern of signal change was not found in either correct-no-reward trials or incorrect trials.
However, the signal changes between response and food delivery in the correct-reward trials with different delayed interval were not observed despite the delay of reward, indicating that the
AIC may play a role in the gustatory processing, but not in the reward anticipation.
response point 200 ms reward delay 500 ms reward delay 40 800 ms reward delay reward omission
incorrect 30
20 (%) 0 F ∆ F/
10
0
-10 -1 -0.50 0.2 0.5 0.8 1 1.5 2 2.5 3 3.5 4 Time (seconds) Figure 17. Response-locked averaged calcium transient in correct-reward trials with delayed reward interval of 200 ms (blue), 500 ms (yellow), and 800 ms (magenta), correct-no-reward trials (green), and incorrect trials (red) across time. The cyan vertical line represents the time point when response was made. The blue, yellow, and magenta vertical lines represent food delivery for each delayed reward, respectively. Grey shaded areas indicate the SEM.
71
40
30
20 (%) 0 F ∆ F/
10
0
-10 -1 -0.50 0.5 1 1.5 2 2.5 3 3.5 4 Time (seconds)
Figure 18. Delivery-locked averaged calcium transient in all correct-reward trials. The cyan vertical line represents the time when response was made, and the purple vertical line represents food delivery. Grey shaded areas indicate the SEM. Discussion
The role of the AIC in cognitive control in the mouse model
State uncertainty processing. The AIC of mice may be implicated in state uncertainty processing. State uncertainty refers to the uncertainty associated with the identity of the current state (Yoshida & Ishii, 2006) and the anticipation of and preparation for impending stimuli (Bach
& Dolan, 2012; Fan et al., 2014). The fiber photometry results showed a slight decrease of signals in the AIC during the 500 ms window after the cue sound that initiated the trial compared to the baseline period (i.e., 1.5 s starting from 0.5 s preceding the onset of the cue sound), suggesting that the AIC is involved in state uncertainty processing. This finding is consistent with human neuroimaging evidence showing the deactivation of the AIC under low uncertainty compared to state uncertainty (i.e., during the fixation period before trial onset; Fan et al., 2014).
Mice successfully established the association between the cue sound and upcoming events so
72
that the state uncertainty during this short period was reduced and the signal of the AIC dropped.
However, the gap between the cue sound and the lever extension was too long (i.e., approximately 1000 ms) so that state uncertainty rose again, and a significant increase of signals occurred during the following window of 500 ms until the levers were extended. Mice also learned the association between the lever extension and approaching behavioral events; therefore, the signal kept dropping until the target was presented. In the optogenetic experiments, unilateral/bilateral inhibition of the AIC after the cue sound resulted in reduced behavioral performance under cognitive control, further supporting the argument that the AIC is associated with state uncertainty processing.
Network global efficiency. The AIC, as a hub of the CCN, may also play an important role in network global efficiency underlying cognitive control. Global efficiency is defined as the global capacity of a network to transmit information through all possible routes (Avena-
Koenigsberger, Misic, & Sporns, 2017). A network hub refers to the key node that has the shortest paths with other nodes in different modules. A recent human study showed that a simulated lesion of the AIC led to a significant decrease of the global efficiency of the CCN and suggested that the behavioral deficits of cognitive control in patients with AIC lesions could be due to the disruption of the AIC in the global communication of the network (T. Wu et al., 2019).
In the current mouse study, the inhibition of the AIC resulted in a significant decrease in accuracy, which may be related to its role in the global efficiency of the CCN. The silence of the
AIC may have affected information transmissions between plentiful pairs of nodes. However, only the performance of the congruent condition, but not the incongruent condition, was significantly impaired when the AIC was silenced, which might be due to the relatively low accuracy (~0.55, slightly higher than chance level) in the incongruent condition even without
73
inhibition. A behavioral task with higher reliability and validity should be designed in future studies. In addition, the small sample size in the mouse study may have impacted the statistical power. Although six animals per group was considered an adequate sample size, the scientific and statistical basis of this notion has been questioned (Charan & Kantharia, 2013). A larger sample size may be needed to further examine the role of the AIC in the mouse model.
Reward-based association learning. A striking increase of calcium-based fluorescence signals was shown starting from 800 ms after the correct response was made and the signals reached the apex after 700 ms, reflecting the potential role of the AIC in reward-based association learning. To achieve goal-directed behaviors, associations need to be learned between stimuli, responses, rules, and corresponding rewards (Ridderinkhof, van den
Wildenberg, Segalowitz, & Carter, 2004). The neural activity of the AIC in mice facilitates the process of cognitive control to make advantageous decisions, suggesting the importance of the
AIC in reward-oriented behaviors (Kusumoto-Yoshida et al., 2015). We could speculate that if the AIC is inhibited during the reward period, the association learning for the paradigm may be impaired and the accuracy may gradually drop to chance level. Moreover, no significant difference in the signal increase after a correct response (i.e., reward) was shown between the congruent and incongruent conditions, suggesting that reward-based association learning may not interact with conflict processing.
In the first experiment with the fiber photometry recording, food was delivered immediately when the correct response was made. Thus, it was difficult to determine whether the signal increase in the AIC was related to the correct response or to food consumption. In the subsequent experiment, correct response and food delivery were separated by delayed intervals.
The results showed that there was no signal change between the correct response and food
74
delivery and the signal started to increase around 800 ms and peaked around 1500 ms after the food was delivered, indicating that the AIC may play a role in gustatory processing. These findings are consistent with the notions that the insular cortex is critical to establishing taste- reinforced choice behavior (Schiff et al., 2018). In addition, human neuroimaging evidence that shows that the activation of the bilateral AIC is associated with reward anticipation but not reward outcome (Liu, Hairston, Schrier, & Fan, 2011). However, preliminary results in the current study showed that omission of rewards did not lead to the rise and fall of signal in the
AIC, indicating that the AIC is not associated with reward anticipation. The neuromechanism of reward anticipation should be further examined with a full sample.
A compensatory role of the hemispheric AIC. The differences of impairment in cognitive control between unilateral and bilateral inhibition of the AIC for both inhibition periods suggest a compensatory role of the contralateral AIC in cognitive processing (Swick et al., 2008). For inhibition after the cue sound, the accuracy of the congruent conditions was reduced for both unilateral and bilateral inhibition of the AIC, and accuracy dropped more for bilateral inhibition than for unilateral inhibition. These results indicate that only one hemispheric
AIC was not adequate to support state uncertainty processing. On the contrary, for inhibition during the presentation of target and distractor stimuli, the accuracy of the congruent condition decreased only when the unilateral AIC was inhibited, indicating that one hemispheric AIC can compensate for the counter-lateral disruption of the AIC by taking over the function of maintaining the global efficiency of the CCN. In future studies, the relationship between cognitive control and the laterality of the AIC can be further examined.
75
Neuroanatomy of the AIC
The connectivity between the AIC and other regions may subserve cognitive control jointly. The AIC has been divided into dorsal (AId), ventral (AIv), and posterior (AIp) parts (van de Werd, Rajkowska, Evers, & Uylings, 2010). The neurons in the AId and AIp project to the bilateral claustrum with more projections to the ipsilateral claustrum than to the contralateral (Q.
Wang et al., 2017), indicating that inputs from the AId and AIp can be processed in the bilateral claustrum. In addition, strong claustrum inputs are directed to the ipsilateral AId, AIv, and AIp, but sparser projections have been found from the claustrum to the contralateral AId and AIp (Q.
Wang et al., 2017), indicating that to a great extent the information exchange between the AId and AIp takes place in the ipsilateral hemisphere. The claustrum has been shown to receive inputs from the ACC for top-down cognitive control (White et al., 2018), which may provide hints about the roles of the AIC and the claustrum in cognitive control. The AIC may also send inputs to the claustrum for top-down cognitive control. Moreover, in line with the connectivity between the ACC and the AIC in the human brain, the AIC of mice is reciprocally connected to the ACC; however, the interconnectivity between the AIC and the claustrum is richer than between the AIC and the ACC (Qadir et al., 2018). In future studies, the contribution of the circuitry of the AIC-claustrum-ACC to cognitive control should be further examined.
Functional organization of the insular cortex in mice remains ambiguous, and the functional roles of the agranular insula, granular insula, and dysgranular insula need to be further differentiated. The agranular regions are phylogenetically more primitive than the granular regions of the prefrontal cortex (Carlén, 2017), but the difference between the agranular insula and the granular insula has been poorly described. An examination of the functional heterogeneity in the insular cortex may advance our understanding of the important structure.
76
Conclusion
The AIC of mice is necessary for state uncertainty processing under cognitive control.
Deficits in cognitive control are core features of psychiatric disorders. The findings in the current mouse model may advance the understanding of the functional role of the AIC and shed some light on future studies to clarify the relationships between the AIC and mental illness.
77
CHAPTER 5
General Discussion
Cognitive control: from mouse to human
The mouse model examining the relationship between cognitive control and the AIC allowed us to refine the functional annotation of the AIC. The human lesion study in this dissertation showed that the AIC is critical for the processing efficiency of cognitive control. The mouse study further revealed that the AIC may have an active role in state uncertainty processing and enhancing reward-based association learning. The contribution of the AIC to cognitive control was identified in the human lesion study, and the functional role of the bilateral AIC in state uncertainty processing under cognitive control was further specified. The current findings uncover important information about the similarity of the role of the AIC in cognitive control between humans and mice and may provide insights for mammalian evolution. The mouse model has advantages to serve as the model organism to study the genes and structures for examining the mechanism underlying dysfunctions and diseases in humans. The mouse model also has the feasibility to conduct invasive manipulations in the brain that can establish the causality between functions and structures and allow us to overcome the limitations in spatial and temporal resolution in human brain imaging studies (Namkung, Kim, & Sawa, 2017).
The AIC, the processing efficiency of cognitive control, and the CCC
The AIC is critical for the processing efficiency of cognitive control, which provides new understanding of the functional role of the AIC in cognitive control. This finding is in line with the notion that the AIC serves as a bottleneck of cognitive control (T. Wu et al., 2019). The activation of the AIC is associated with the CCC, and the disruption in the AIC leads to reduction in the CCC, suggesting that the upper limit of information that can be reliably
78
processed within time limits via cognitive control is constrained by the function of the AIC (T.
Wu et al., 2019). The current dissertation showed a decrease in processing efficiency as a consequence of the disruption of the AIC, indicating that the speed of information processing via cognitive control was also restricted by the function of the AIC. The AIC is essential for cognitive control, in terms of both capacity and processing efficiency. In addition, one of our previous studies showed that the conflict effect of RT was correlated with the CCC without taking accuracy into account (Chen et al., 2019). Hence, we speculated that the CCC should be correlated with the processing efficiency of cognitive control. The relationships between the
AIC, the CCC, and the processing efficiency of cognitive control can be further examined in future studies to contribute to a comprehensive view of cognitive control.
The AIC and uncertainty processing
As shown by prior evidence in the human study, the AIC is involved in state uncertainty processing. The AIC was deactivated when task uncertainty was lower than state uncertainty whereas it was activated when task uncertainty was higher than state uncertainty (Fan et al.,
2014). Although a small heave was detected starting from the offset of target presentation to distractor presentation in the current study, it was far less than the signal increase during the second 500 ms window before the lever extension, indicating that uncertainty about the distractor
(i.e., task uncertainty) was much lower than state uncertainty before the lever extension. In future mouse studies, a behavioral task with much higher computational task uncertainty needs to be developed to further identify the role of the AIC under cognitive control. In addition, human neuroimaging evidence has shown that the AIC activates as a function of information uncertainty
(Fan et al., 2014). Uncertainty in the incongruent condition is higher than in the congruent condition, which should result in significant difference of signals in the AIC between the
79
congruent and incongruent conditions. However, the signal changes of the AIC in the congruent condition in the current mouse study were not different from the incongruent condition, indicating that the difference of uncertainty levels between these two conditions may not have been adequate to cause distinct signal patterns. Different levels of uncertainty should be manipulated to advance the understanding of the relationship between the AIC and uncertainty processing in future studies.
The AIC and reward-based association learning
The AIC is implicated in reward-based association learning in the current mouse model, supporting a role of the AIC in encoding incentive values of stimuli to mediate cognitive behaviors. Rewards are accompanied by feelings of pleasure, which in turn affect the conscious decision-making process for particular actions (Namkung et al., 2017). Converging evidence suggests that the AIC encodes the association between rewarding stimuli and subjective feeling states and initiates cognitive processes to further process the salient inputs (Menon & Uddin,
2010; Uddin, 2015). This notion regarding the roles of the AIC in cognitive control and reward- related processing was also supported by the marked calcium-based signal change after reward delivery in the mouse study of this dissertation. In addition, although no aversive processing was involved in the current mouse study, the AIC has been shown to be also engaged in mediating cognitive behaviors based on aversive feelings such as pain (Dolan, 2002). In addition to its role in reward-based association learning, the AIC also plays a critical role in punishment-based avoidance learning (Palminteri et al., 2012). It is recommended that future studies direct more attention to the relationships between the AIC, feeling states, and cognitive control.
80
The AIC, cognitive control, and higher level cognition
An examination of cognitive control and the functional roles of central structures underlying cognitive control demand attention to understanding the fundamental aspects of human cognition. Cognitive control serves as a core process underlying broadly defined executive functions and higher level cognition such as intelligence, especially fluid intelligence
(Chen et al., 2019). Given that the AIC serves as one of the key nodes for cognitive control, especially for its capacity and processing efficiency, future studies should focus on the relationships between the AIC, cognitive control, and higher order cognition. A multiple-demand system has been identified in a wide variety of cognitive tasks and in standard tests of fluid intelligence, including regions in the prefrontal and parietal cortices; these include the inferior frontal sulcus, the AIC and adjacent frontal operculum, pre-supplementary motor area and adjacent dorsal ACC, and the IPS (Duncan, 2010; Duncan et al., 2000; Prabhakaran, Smith,
Desmond, Glover, & Gabrieli, 1997), which overlap substantially with the CCN. The relationship between cognitive control and fluid intelligence may be mediated by the CCN and especially by the AIC due to its necessity in the capacity and processing efficiency of cognitive control as a bottleneck.
A functional architecture of cognitive control
Diverse regions in the CCN supporting cognitive control may have segregated functions.
Cognitive control, a fundamental process involved in flexible coordination of thoughts and behaviors in response to the dynamic interplay between external inputs and internal goals, is associated with the activity of the frontoparietal network, including regions such as the FEF and
IPS, and the cingulo-opercular network, including regions such as the AIC and the ACC, in previous functional neuroimaging studies (Dosenbach et al., 2007; Fan et al., 2014; T. Wu et al.,
81
2018). The AIC is essential for the capacity of cognitive control (T. Wu et al., 2019), the processing efficiency of cognitive control demonstrated in the current lesion study, and state uncertainty processing in the current mouse study. The ACC, on the contrary, may not be necessary for the capacity of cognitive control (T. Wu et al., 2019) or the processing efficiency of cognitive control. As a limbic motor region, the ACC projects densely to the motor cortex and spinal cord (Paus, 2001), indicating that the ACC may be responsible for the action-end of cognitive control to guide motor response. The ACC is also implicated in error detection (Carter et al., 1998), providing hints for its role in performance monitoring. In addition, the FEF and the
IPS contribute greatly to exerting top-down cognitive control on bottom-up sensory areas such as the visual cortex (Bressler, Tang, Sylvester, Shulman, & Corbetta, 2008; Zhou & Desimone,
2011), and the IPS has been shown to be sensitive to the salience of stimuli (Geng & Mangun,
2009). A potential functional architecture of cognitive control is proposed here: 1) the FEF and the IPS may play roles in preliminary information processing on the inputs for sensory areas in different modalities, 2) higher level regions in the CCN such as the AIC may further abstract and integrate preliminary processed information, and 3) the ACC may mainly participate in motor responses based on abstract information from the AIC and error detection to send feedback to the
AIC. If the information to be processed in the AIC exceeds the capacity of cognitive control, biased information for response may be passed to the ACC, which leads to errors. The functional architecture of cognitive control should be established and tested by the examination of functional heterogeneity in the CCN.
Conclusion
A causal role of the AIC in cognitive control in terms of processing efficiency was demonstrated in the lesion study. A translational paradigm was developed to measure cognitive
82
control in mice, and a critical role of the AIC of both hemispheres in state uncertainty processing was further demonstrated using the techniques of optogenetics and fiber photometry in the subsequent mouse study. Combining the lesion-based evidence and the mouse model revealed important insights to understand the functional role of the AIC in cognitive control at both the circuit level and the cell level. The methods and findings in the current study can provide novel insight into subsequent empirical work to examine the functional specialization of different regions in the CCN. The current studies can also provide supplementary functional implications for the roles of the AIC in clinical populations, especially for individuals with cognitive disorders.
83
References
Accolla, R., & Carleton, A. (2008). Internal body state influences topographical plasticity of
sensory representations in the rat gustatory cortex. Proceedings of the National Academy
of Sciences, 105(10), 4010-4015.
Allman, J. M., Tetreault, N. A., Hakeem, A. Y., Manaye, K. F., Semendeferi, K., Erwin, J. M., . .
. Hof, P. R. (2011). The von Economo neurons in fronto-insular and anterior cingulate
cortex. Annals of the New York Academy of Sciences, 1225, 59-71.
Anderson, K., Bones, B., Robinson, B., Hass, C., Lee, H., Ford, K., . . . Jacobs, B. (2009). The
morphology of supragranular pyramidal neurons in the human insular cortex: A
quantitative Golgi study. Cerebral Cortex, 19(9), 2131-2144.
Anderson, M., & Green, C. (2001). Suppressing unwanted memories by executive control.
Nature, 410(6826), 366-369. doi:10.1038/35066572
Aston-Jones, G., & Deisseroth, K. (2013). Recent advances in optogenetics and
pharmacogenetics. Brain research, 1511, 1-5. doi:10.1016/j.brainres.2013.01.026
Atlan, G., Terem, A., Peretz-Rivlin, N., Sehrawat, K., Gonzales, B. J., Pozner, G., . . . Citri, A.
(2018). The Claustrum Supports Resilience to Distraction. Current Biology, 28(17),
2752-2762.e2757. doi:10.1016/j.cub.2018.06.068
Augustine, J. R. (1996). Circuitry and functional aspects of the insular lobe in primates including
humans. Brain research reviews, 22(3), 229-244.
Avena-Koenigsberger, A., Misic, B., & Sporns, O. (2017). Communication dynamics in complex
brain networks. Nature reviews neuroscience, 19(1), 17-33. doi:10.1038/nrn.2017.149
84
Avery, J. A., Kerr, K. L., Ingeholm, J. E., Burrows, K., Bodurka, J., & Simmons, W. K. (2015).
A common gustatory and interoceptive representation in the human mid‐insula. Human
brain mapping, 36(8), 2996-3006.
Bach, D. R., & Dolan, R. J. (2012). Knowing how much you don't know: a neural organization
of uncertainty estimates. Nature reviews neuroscience, 13(8), 572-586.
doi:10.1038/nrn3289
Bamiou, D.-E., Musiek, F. E., & Luxon, L. M. (2003). The insula (Island of Reil) and its role in
auditory processing: Literature review. Brain research reviews, 42(2), 143-154.
Bar-On, R. (2003). Exploring the neurological substrate of emotional and social intelligence.
Brain, 126(8), 1790-1800. doi:10.1093/brain/awg177
Baumeister, R. F., Vohs, K. D., & Tice, D. M. (2007). The strength model of self-control.
Current directions in psychological science, 16(6), 351-355.
Binder, D. K., Schaller, K., & Clusmann, H. (2007). The seminal contributions of Johann-
Christian Reil to anatomy, physiology, and psychiatry. Neurosurgery, 61(5), 1091-1096.
Botvinick, M., Nystrom, L. E., Fissell, K., Carter, C. S., & Cohen, J. D. (1999). Conflict
monitoring versus selection-for-action in anterior cingulate cortex. Nature, 402(6758),
179.
Brass, M., & Haggard, P. (2007). To Do or Not to Do: The Neural Signature of Self-Control.
Journal of Neuroscience, 27(34), 9141-9145. doi:10.1523/jneurosci.0924-07.2007
Bressler, S. L., Tang, W., Sylvester, C. M., Shulman, G. L., & Corbetta, M. (2008). Top-down
control of human visual cortex by frontal and parietal cortex in anticipatory visual spatial
attention. Journal of Neuroscience, 28(40), 10056-10061.
85
Butti, C., & Hof, P. R. (2010). The insular cortex: A comparative perspective. Brain Structure
and Function, 214(5-6), 477-493. doi:10.1007/s00429-010-0264-y
Cai, W., Ryali, S., Chen, T., Li, C. S. R., & Menon, V. (2014). Dissociable Roles of Right
Inferior Frontal Cortex and Anterior Insula in Inhibitory Control: Evidence from Intrinsic
and Task-Related Functional Parcellation, Connectivity, and Response Profile Analyses
across Multiple Datasets. Journal of Neuroscience, 34(44), 14652-14667.
doi:10.1523/jneurosci.3048-14.2014
Carlén, M. (2017). What constitutes the prefrontal cortex? Science, 358(6362), 478-482.
Carter, C. S., Braver, T. S., Barch, D. M., Botvinick, M. M., Noll, D., & Cohen, J. D. (1998).
Anterior cingulate cortex, error detection, and the online monitoring of performance.
Science, 280(5364), 747-749.
Cauda, F., D'Agata, F., Sacco, K., Duca, S., Geminiani, G., & Vercelli, A. (2011). Functional
connectivity of the insula in the resting brain. Neuroimage, 55(1), 8-23.
doi:10.1016/j.neuroimage.2010.11.049
Chand, G. B., & Dhamala, M. (2017). Interactions between the anterior cingulate-insula network
and the fronto-parietal network during perceptual decision-making. Neuroimage, 152,
381-389. doi:10.1016/j.neuroimage.2017.03.014
Charan, J., & Kantharia, N. (2013). How to calculate sample size in animal studies? Journal of
pharmacology & pharmacotherapeutics, 4(4), 303-306.
Chen, A., Oathes, D. J., Chang, C., Bradley, T., Zhou, Z.-W., Williams, L. M., . . . Etkin, A.
(2013). Causal interactions between fronto-parietal central executive and default-mode
networks in humans. Proceedings of the National Academy of Sciences of the United
States of America, 110(49), 19944-19949.
86
Chen, T.-W., Wardill, T. J., Sun, Y., Pulver, S. R., Renninger, S. L., Baohan, A., . . . Jayaraman,
V. (2013). Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature,
499(7458), 295-300.
Chen, Y., Chen, C., Wu, T., Qiu, B., Zhang, W., & Fan, J. (2020). Accessing the Development
and Heritability of the Capacity of Cognitive Control. Neuropsychologia, 139, 107361.
Chen, Y., Spagna, A., Wu, T., Kim, T. H., Wu, Q., Chen, C., . . . Fan, J. (2019). Testing a
Cognitive Control Model of Human Intelligence. Scientific Reports, 9(1), 2898.
doi:10.1038/s41598-019-39685-2
Chiu, Y.-C., & Yantis, S. (2009). A domain-independent source of cognitive control for task
sets: Shifting spatial attention and switching categorization rules. Journal of
Neuroscience, 29(12), 3930-3938.
Cloutman, L. L., Binney, R. J., Drakesmith, M., Parker, G. J., & Lambon Ralph, M. A. (2012).
The variation of function across the human insula mirrors its patterns of structural
connectivity: evidence from in vivo probabilistic tractography. Neuroimage, 59(4), 3514-
3521. doi:10.1016/j.neuroimage.2011.11.016
Cockrell, J. R., & Folstein, M. F. (2002). Mini-mental state examination. In J. R. M. Copeland,
M. T. Abou-Saleh, & D. G. Blazer (Eds.), Principles and practice of geriatric psychiatry
(2nd ed., pp. 140-141). New York, NY: Wiley & Sons.
Cohen, R., Kaplan, R., Moser, D., Jenkins, M., & Wilkinson, H. (1999). Impairments of attention
after cingulotomy. Neurology, 53(4), 819-824.
Cole, M. W., & Schneider, W. (2007). The cognitive control network: Integrated cortical regions
with dissociable functions. Neuroimage, 37(1), 343-360.
doi:10.1016/j.neuroimage.2007.03.071
87
Corbetta, M. (1998). Frontoparietal cortical networks for directing attention and the eye to visual
locations: Identical, independent, or overlapping neural systems? Proceedings of the
National Academy of Sciences of the United States of America, 95(3), 831-838.
Corbetta, M., Kincade, J. M., Ollinger, J. M., McAvoy, M. P., & Shulman, G. L. (2000).
Voluntary orienting is dissociated from target detection in human posterior parietal
cortex. Nature neuroscience, 3(3), 292-297.
Corbetta, M., & Shulman, G. L. (2002). Control of goal-directed and stimulus-driven attention in
the brain. Nature reviews neuroscience, 3(3), 201-215.
Craig, A. (2009). How do you feel--now? The anterior insula and human awareness. Nature
reviews neuroscience, 10(1), 59-70.
Crick, F. C., & Koch, C. (2005). What is the function of the claustrum? Philosophical
Transactions of the Royal Society of London. Series B, Biologial Sciences, 360(1458),
1271-1279. doi:10.1098/rstb.2005.1661
Critchley, H. D., Wiens, S., Rotshtein, P., Ohman, A., & Dolan, R. J. (2004). Neural systems
supporting interoceptive awareness. Nature neuroscience, 7(2), 189-195.
doi:10.1038/nn1176
Dalley, J. W., Everitt, B. J., & Robbins, T. W. (2011). Impulsivity, compulsivity, and top-down
cognitive control. Neuron, 69(4), 680-694. doi:10.1016/j.neuron.2011.01.020
Deisseroth, K. (2011). Optogenetics. Nature Methods, 8(1), 26-29. doi:10.1038/nmeth.f.324
Derakshan, N., & Eysenck, M. W. (2009). Anxiety, processing efficiency, and cognitive
performance: New developments from attentional control theory. European Psychologist,
14(2), 168-176.
88
Devue, C., Collette, F., Balteau, E., Degueldre, C., Luxen, A., Maquet, P., & Brédart, S. (2007).
Here I am: The cortical correlates of visual self-recognition. Brain research, 1143, 169-
182.
Di Pellegrino, G., Ciaramelli, E., & Làdavas, E. (2007). The regulation of cognitive control
following rostral anterior cingulate cortex lesion in humans. Journal of cognitive
neuroscience, 19(2), 275-286.
Diamond, A. (2013). Executive functions. Annual Review of Psychology, 64, 135-168.
doi:10.1146/annurev-psych-113011-143750
Dolan, R. J. (2002). Emotion, cognition, and behavior. Science, 298(5596), 1191-1194.
Dosenbach, N. U., Fair, D. A., Miezin, F. M., Cohen, A. L., Wenger, K. K., Dosenbach, R. A., . .
. Petersen, S. E. (2007). Distinct brain networks for adaptive and stable task control in
humans. Proceedings of the National Academy of Sciences of the United States of
America, 104(26), 11073-11078. doi:10.1073/pnas.0704320104
Dosenbach, N. U., Visscher, K. M., Palmer, E. D., Miezin, F. M., Wenger, K. K., Kang, H. C., . .
. Petersen, S. E. (2006). A core system for the implementation of task sets. Neuron, 50(5),
799-812. doi:10.1016/j.neuron.2006.04.031
Duckworth, A. L. (2011). The significance of self-control. Proceedings of the National Academy
of Sciences of the United States of America, 108(7), 2639-2640.
Duncan, J. (2010). The multiple-demand (MD) system of the primate brain: Mental programs for
intelligent behaviour. Trends in cognitive sciences, 14(4), 172-179.
doi:10.1016/j.tics.2010.01.004
Duncan, J., Seitz, R. J., Kolodny, J., Bor, D., Herzog, H., Ahmed, A., . . . Emslie, H. (2000). A
neural basis for general intelligence. Science, 289(5478), 457-460.
89
Eagle, D. M., & Robbins, T. W. (2003). Inhibitory Control in Rats Performing a Stop-Signal
Reaction-Time Task: Effects of Lesions of the Medial Striatum and d-Amphetamine.
Behavioral Neuroscience, 117(6), 1302-1317. doi:10.1037/0735-7044.117.6.1302
Eckert, U., Metzger, C. D., Buchmann, J. E., Kaufmann, J., Osoba, A., Li, M., . . . Walter, M.
(2012). Preferential networks of the mediodorsal nucleus and centromedian-
parafascicular complex of the thalamus--a DTI tractography study. Human brain
mapping, 33(11), 2627-2637. doi:10.1002/hbm.21389
Fan, J. (2014). An information theory account of cognitive control. Frontiers in human
neuroscience, 8, 680.
Fan, J., McCandliss, B. D., Fossella, J., Flombaum, J. I., & Posner, M. I. (2005). The activation
of attentional networks. Neuroimage, 26(2), 471-479.
doi:10.1016/j.neuroimage.2005.02.004
Fan, J., McCandliss, B. D., Sommer, T., Raz, A., & Posner, M. I. (2002). Testing the efficiency
and independence of attentional networks. Journal of cognitive neuroscience, 14(3), 340-
347.
Fan, J., Van Dam, N. T., Gu, X., Liu, X., Wang, H., Tang, C. Y., & Hof, P. R. (2014).
Quantitative characterization of functional anatomical contributions to cognitive control
under uncertainty. Journal of cognitive neuroscience, 26(7), 1490-1506.
Farrer, C., & Frith, C. D. (2002). Experiencing oneself vs another person as being the cause of an
action: The neural correlates of the experience of agency. Neuroimage, 15(3), 596-603.
Feinstein, J. S., Khalsa, S. S., Salomons, T. V., Prkachin, K. M., Frey-Law, L. A., Lee, J. E., . . .
Rudrauf, D. (2016). Preserved emotional awareness of pain in a patient with extensive
90
bilateral damage to the insula, anterior cingulate, and amygdala. Brain Structure and
Function, 221(3), 1499-1511. doi:10.1007/s00429-014-0986-3
Fellows, L. K., & Farah, M. J. (2005). Is anterior cingulate cortex necessary for cognitive
control? Brain, 128(4), 788-796.
Fenno, L., Yizhar, O., & Deisseroth, K. (2011). The development and application of
optogenetics. Annual Review of Neuroscience, 34, 389-412. doi:10.1146/annurev-neuro-
061010-113817
Flynn, F. G. (1999). Anatomy of the insula functional and clinical correlates. Aphasiology, 13(1),
55-78. doi:10.1080/026870399402325
Fox, M. D., Corbetta, M., Snyder, A. Z., Vincent, J. L., & Raichle, M. E. (2006). Spontaneous
neuronal activity distinguishes human dorsal and ventral attention systems. Proceedings
of the National Academy of Sciences of the United States of America, 103(26), 10046-
10051.
Fujita, K. (2011). On conceptualizing self-control as more than the effortful inhibition of
impulses. Personality and Social Psychology Review, 15(4), 352-366.
Fukuda, K., Awh, E., & Vogel, E. K. (2010). Discrete capacity limits in visual working memory.
Current Opinion in Neurobiology, 20(2), 177-182. doi:10.1016/j.conb.2010.03.005
Gallay, D. S., Gallay, M., Jeanmonod, D., Rouiller, E. M., & Morel, A. (2011). The insula of
Reil revisited: Multiarchitectonic organization in macaque monkeys. Cerebral Cortex,
22(1), 175-190.
Geng, J. J., & Mangun, G. R. (2009). Anterior intraparietal sulcus is sensitive to bottom–up
attention driven by stimulus salience. Journal of cognitive neuroscience, 21(8), 1584-
1601.
91
Ghaziri, J., Tucholka, A., Girard, G., Houde, J. C., Boucher, O., Gilbert, G., . . . Nguyen, D. K.
(2017). The Corticocortical Structural Connectivity of the Human Insula. Cerebral
Cortex, 27(2), 1216-1228. doi:10.1093/cercor/bhv308
Gogolla, N. (2017). The insular cortex. Current Biology, 27(12), R580-R586.
Gogolla, N., Takesian, A. E., Feng, G., Fagiolini, M., & Hensch, T. K. (2014). Sensory
integration in mouse insular cortex reflects GABA circuit maturation. Neuron, 83(4),
894-905. doi:10.1016/j.neuron.2014.06.033
Goulden, N., Khusnulina, A., Davis, N. J., Bracewell, R. M., Bokde, A. L., McNulty, J. P., &
Mullins, P. G. (2014). The salience network is responsible for switching between the
default mode network and the central executive network: Replication from DCM.
Neuroimage, 99, 180-190.
Greicius, M. D., Krasnow, B., Reiss, A. L., & Menon, V. (2003). Functional connectivity in the
resting brain: A network analysis of the default mode hypothesis. Proceedings of the
National Academy of Sciences of the United States of America, 100(1), 253-258.
Gu, X., Gao, Z., Wang, X., Liu, X., Knight, R. T., Hof, P. R., & Fan, J. (2012). Anterior insular
cortex is necessary for empathetic pain perception. Brain, 135(9), 2726-2735.
Gu, X., Hof, P. R., Friston, K. J., & Fan, J. (2013). Anterior insular cortex and emotional
awareness. Journal of Comparative Neurology, 521(15), 3371-3388.
doi:10.1002/cne.23368
Hahn, B., Ross, T. J., & Stein, E. A. (2006). Neuroanatomical dissociation between bottom-up
and top-down processes of visuospatial selective attention. Neuroimage, 32(2), 842-853.
doi:10.1016/j.neuroimage.2006.04.177
92
Halassa, M. M., Chen, Z., Wimmer, R. D., Brunetti, P. M., Zhao, S., Zikopoulos, B., . . . Wilson,
M. A. (2014). State-dependent architecture of thalamic reticular subnetworks. Cell,
158(4), 808-821. doi:10.1016/j.cell.2014.06.025
Hasson, U., Avidan, G., Deouell, L. Y., Bentin, S., & Malach, R. (2003). Face-selective
activation in a congenital prosopagnosic subject. Journal of cognitive neuroscience,
15(3), 419-431.
He, H., Xu, P., Wu, T., Chen, Y., Wang, J., Qiu, Y., . . . Luo, Y. (2019). Reduced Capacity of
Cognitive Control in Older Adults with Mild Cognitive Impairment. Journal of
Alzheimer's Disease, 71(1), 185-200.
Heining, M., Young, A. W., Ioannou, G., Andrew, C. M., Brammer, M. J., Gray, J. A., &
Phillips, M. L. (2003). Disgusting smells activate human anterior insula and ventral
striatum. Annals of the New York Academy of Sciences, 1000(1), 380-384.
Heitz, R. P. (2014). The speed-accuracy tradeoff: History, physiology, methodology, and
behavior. Frontiers in Neuroscience, 8, 150. doi:10.3389/fnins.2014.00150
Jakab, A., Molnár, P. P., Bogner, P., Béres, M., & Berényi, E. L. (2012). Connectivity-based
parcellation reveals interhemispheric differences in the insula. Brain topography, 25(3),
264-271.
Janer, K. W., & Pardo, J. V. (1991). Deficits in selective attention following bilateral anterior
cingulotomy. Journal of cognitive neuroscience, 3(3), 231-241.
Katz, D. B., Simon, S., & Nicolelis, M. A. (2001). Dynamic and multimodal responses of
gustatory cortical neurons in awake rats. Journal of Neuroscience, 21(12), 4478-4489.
93
Kiehl, K. A., Hare, R. D., Liddle, P. F., & McDonald, J. J. (1999). Reduced P300 responses in
criminal psychopaths during a visual oddball task. Biological psychiatry, 45(11), 1498-
1507.
Kim, D. S., Jayaraman, V., Looger, L. L., & Svoboda, K. (2014). Engineering fluorescent
calcium sensor proteins for imaging neural activity. Society for Neuroscience, 11-19.
Koike, H., Demars, M. P., Short, J. A., Nabel, E. M., Akbarian, S., Baxter, M. G., & Morishita,
H. (2016). Chemogenetic Inactivation of Dorsal Anterior Cingulate Cortex Neurons
Disrupts Attentional Behavior in Mouse. Neuropsychopharmacology, 41(4), 1014-1023.
doi:10.1038/npp.2015.229
Kosillo, P., & Smith, A. T. (2010). The role of the human anterior insular cortex in time
processing. Brain Structure and Function, 214(5-6), 623-628. doi:10.1007/s00429-010-
0267-8
Koziol, L. F. (2014). Cognitive control, reward, and the basal ganglia The myth of executive
functioning (pp. 61-64). New York, NY: Springer.
Kusumoto-Yoshida, I., Liu, H., Chen, B. T., Fontanini, A., & Bonci, A. (2015). Central role for
the insular cortex in mediating conditioned responses to anticipatory cues. Proceedings of
the National Academy of Sciences of the United States of America, 112(4), 1190-1195.
doi:10.1073/pnas.1416573112
Lawton‐Craddock, A., Nixon, S. J., & Tivis, R. (2003). Cognitive efficiency in stimulant
abusers with and without alcohol dependence. Alcoholism: Clinical and Experimental
Research, 27(3), 457-464.
94
Li, Y., Zhong, W., Wang, D., Feng, Q., Liu, Z., Zhou, J., . . . Luo, M. (2016). Serotonin neurons
in the dorsal raphe nucleus encode reward signals. Nature Communications, 7, 10503.
doi:10.1038/ncomms10503
Liang, Z., Ma, Y., Watson, G. D. R., & Zhang, N. (2017). Simultaneous GCaMP6-based fiber
photometry and fMRI in rats. Journal of Neuroscience Methods, 289, 31-38.
doi:10.1016/j.jneumeth.2017.07.002
Liu, X., Hairston, J., Schrier, M., & Fan, J. (2011). Common and distinct networks underlying
reward valence and processing stages: A meta-analysis of functional neuroimaging
studies. Neuroscience & Biobehavioral Reviews, 35(5), 1219-1236.
doi:10.1016/j.neubiorev.2010.12.012
Logan, G. D. (1994). On the ability to inhibit thought and action: A users' guide to the stop
signal paradigm. In D. Dagenbach & T. H. Carr (Eds.), Inhibitory processes in attention,
memory and language (pp. 189-239). San Diego, CA: Academic Press.
Logan, G. D., & Cowan, W. B. (1984). On the ability to inhibit thought and action: A theory of
an act of control. Psychological review, 91(3), 295.
MacDonald, A. W., Cohen, J. D., Stenger, V. A., & Carter, C. S. (2000). Dissociating the role of
the dorsolateral prefrontal and anterior cingulate cortex in cognitive control. Science,
288(5472), 1835-1838.
Mackie, M.-A., Van Dam, N. T., & Fan, J. (2013). Cognitive control and attentional functions.
Brain and cognition, 82(3), 301-312.
Maffei, A., Haley, M., & Fontanini, A. (2012). Neural processing of gustatory information in
insular circuits. Current Opinions in Neurobiology, 22(4), 709-716.
doi:10.1016/j.conb.2012.04.001
95
Mahn, M., Gibor, L., Patil, P., Cohen-Kashi Malina, K., Oring, S., Printz, Y., . . . Yizhar, O.
(2018). High-efficiency optogenetic silencing with soma-targeted anion-conducting
channelrhodopsins. Nature Communications, 9(1), 4125. doi:10.1038/s41467-018-06511-
8
Manoliu, A., Riedl, V., Zherdin, A., Mühlau, M., Schwerthöffer, D., Scherr, M., . . . Bäuml, J.
(2013). Aberrant dependence of default mode/central executive network interactions on
anterior insular salience network activity in schizophrenia. Schizophrenia bulletin, 40(2),
428-437.
Markostamou, I., Rudolf, J., Tsiptsios, I., & Kosmidis, M. H. (2015). Impaired executive
functioning after left anterior insular stroke: A case report. Neurocase, 21(2), 148-153.
doi:10.1080/13554794.2013.878725
Martin, R. C., Loring, D. W., Meador, K. J., & Lee, G. P. (1990). The effects of lateralized
temporal lobe dysfunction on normal and semantic word fluency. Neuropsychologia,
28(8), 823-829.
Mazzola, L., Isnard, J., & Mauguiere, F. (2005). Somatosensory and pain responses to
stimulation of the second somatosensory area (SII) in humans. A comparison with SI and
insular responses. Cerebral Cortex, 16(7), 960-968.
Menon, V., Adleman, N. E., White, C. D., Glover, G. H., & Reiss, A. L. (2001). Error‐related
brain activation during a Go/NoGo response inhibition task. Human brain mapping,
12(3), 131-143.
Menon, V., & Uddin, L. Q. (2010). Saliency, switching, attention and control: A network model
of insula function. Brain Structure and Function, 214(5-6), 655-667.
doi:10.1007/s00429-010-0262-0
96
Mesulam, M. M., & Mufson, E. J. (1982). Insula of the old world monkey. III: Efferent cortical
output and comments on function. Journal of Comparative Neurology, 212(1), 38-52.
Mesulam, M. M., & Mufson, E. J. (1985). The insula of Reil in man and monkey:
Architectonics, connectivity and function. In A. Peters & E. O. Jones (Eds.), Association
and auditory cortices (pp. 179-226). New York, NY: Plenum.
Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D.
(2000). The unity and diversity of executive functions and their contributions to complex
"frontal lobe" tasks: A latent variable analysis. Cognitive psychology, 41(1), 49-100.
doi:10.1006/cogp.1999.0734
Mooney, C. Z., & Duval, R. D. (1993). Bootstrapping: A nonparametric approach to statistical
inference. Newbury Park, CA: Sage.
Moran, L. V., Tagamets, M. A., Sampath, H., O’Donnell, A., Stein, E. A., Kochunov, P., &
Hong, L. E. (2013). Disruption of anterior insula modulation of large-scale brain
networks in schizophrenia. Biological psychiatry, 74(6), 467-474.
Naghavi, H. R., Eriksson, J., Larsson, A., & Nyberg, L. (2007). The claustrum/insula region
integrates conceptually related sounds and pictures. Neuroscience letters, 422(1), 77-80.
Namkung, H., Kim, S. H., & Sawa, A. (2017). The Insula: An Underestimated Brain Area in
Clinical Neuroscience, Psychiatry, and Neurology. Trends in neurosciences, 40(4), 200-
207. doi:10.1016/j.tins.2017.02.002
Nelson, S. M., Dosenbach, N. U., Cohen, A. L., Wheeler, M. E., Schlaggar, B. L., & Petersen, S.
E. (2010). Role of the anterior insula in task-level control and focal attention. Brain
Structure and Function, 214(5-6), 669-680.
97
Nieuwenhuis, S., Yeung, N., van den Wildenberg, W., & Ridderinkhof, K. R. (2003).
Electrophysiological correlates of anterior cingulate function in a go/no-go task: Effects
of response conflict and trial type frequency. Cognitive, affective, & behavioral
neuroscience, 3(1), 17-26.
Nigg, J. T. (2017). Annual Research Review: On the relations among self-regulation, self-
control, executive functioning, effortful control, cognitive control, impulsivity, risk-
taking, and inhibition for developmental psychopathology. Journal of Child Psychology
and Psychiatry, 58(4), 361-383. doi:10.1111/jcpp.12675
Nixon, S. J., Paul, R., & Phillips, M. (1998). Cognitive efficiency in alcoholics and
polysubstance abusers. Alcoholism: Clinical and Experimental Research, 22(7), 1414-
1420.
Ohara, P. T., Granato, A., Moallem, T. M., Wang, B.-R., Tillet, Y., & Jasmin, L. (2003).
Dopaminergic input to GABAergic neurons in the rostral agranular insular cortex of the
rat. Journal of neurocytology, 32(2), 131-141.
Oliveira-Maia, A. J., De Araujo, I. E., Monteiro, C., Workman, V., Galhardo, V., & Nicolelis, M.
A. (2012). The insular cortex controls food preferences independently of taste receptor
signaling. Frontiers in systems neuroscience, 6, 5.
Otis, J. M., Namboodiri, V. M., Matan, A. M., Voets, E. S., Mohorn, E. P., Kosyk, O., . . .
Stuber, G. D. (2017). Prefrontal cortex output circuits guide reward seeking through
divergent cue encoding. Nature, 543(7643), 103-107. doi:10.1038/nature21376
Palminteri, S., Justo, D., Jauffret, C., Pavlicek, B., Dauta, A., Delmaire, C., . . . Durr, A. (2012).
Critical roles for anterior insula and dorsal striatum in punishment-based avoidance
learning. Neuron, 76(5), 998-1009.
98
Papaleo, F., Erickson, L., Liu, G., Chen, J., & Weinberger, D. R. (2012). Effects of sex and
COMT genotype on environmentally modulated cognitive control in mice. Proceedings
of the National Academy of Sciences, 109(49), 20160-20165.
Paus, T. (2001). Primate anterior cingulate cortex: Where motor control, drive and cognition
interface. Nature reviews neuroscience, 2(6), 417-424.
Peretz, I., & Zatorre, R. J. (2005). Brain organization for music processing. Annual Review of
Psychology, 56, 89-114. doi:10.1146/annurev.psych.56.091103.070225
Platel, H., Price, C., Baron, J.-C., Wise, R., Lambert, J., Frackowiak, R., . . . Eustache, F. (1997).
The structural components of music perception. A functional anatomical study. Brain: a
journal of neurology, 120(2), 229-243.
Posner, M. I., & Boies, S. J. (1971). Components of attention. Psychological review, 78(5), 391-
408.
Power, J. D., Cohen, A. L., Nelson, S. M., Wig, G. S., Barnes, K. A., Church, J. A., . . .
Schlaggar, B. L. (2011). Functional network organization of the human brain. Neuron,
72(4), 665-678.
Prabhakaran, V., Smith, J. A., Desmond, J. E., Glover, G. H., & Gabrieli, J. D. (1997). Neural
substrates of fluid reasoning: An fMRI study of neocortical activation during
performance of the Raven's Progressive Matrices Test. Cognitive psychology, 33(1), 43-
63.
Preuschoff, K., Quartz, S. R., & Bossaerts, P. (2008). Human insula activation reflects risk
prediction errors as well as risk. Journal of Neuroscience, 28(11), 2745-2752.
doi:10.1523/JNEUROSCI.4286-07.2008
99
Qadir, H., Krimmel, S. R., Mu, C., Poulopoulos, A., Seminowicz, D. A., & Mathur, B. N.
(2018). Structural Connectivity of the Anterior Cingulate Cortex, Claustrum, and the
Anterior Insula of the Mouse. Frontiers in Neuroanatomy, 12, 100.
doi:10.3389/fnana.2018.00100
Ramautar, J. R., Slagter, H. A., Kok, A., & Ridderinkhof, K. R. (2006). Probability effects in the
stop-signal paradigm: The insula and the significance of failed inhibition. Brain research,
1105(1), 143-154. doi:10.1016/j.brainres.2006.02.091
Reil, J. (1809). Unterfuchungen über den Bau des grofsen Gehirns imMenfchen: Vierte
Fortsetzung VIII. Archiv für die Physiologie Halle, 9, 136-146.
Ridderinkhof, K. R., van den Wildenberg, W. P., Segalowitz, S. J., & Carter, C. S. (2004).
Neurocognitive mechanisms of cognitive control: The role of prefrontal cortex in action
selection, response inhibition, performance monitoring, and reward-based learning. Brain
and cognition, 56(2), 129-140. doi:10.1016/j.bandc.2004.09.016
Rinne, P., Hassan, M., Goniotakis, D., Chohan, K., Sharma, P., Langdon, D., . . . Bentley, P.
(2013). Triple dissociation of attention networks in stroke according to lesion location.
Neurology, 81(9), 812-820.
Rosenthal, N., & Brown, S. (2007). The mouse ascending: Perspectives for human-disease
models. Nature cell biology, 9(9), 993-999.
Rossi, A. F., Pessoa, L., Desimone, R., & Ungerleider, L. G. (2009). The prefrontal cortex and
the executive control of attention. Experimental brain research, 192(3), 489-497.
Rothbart, M. K., & Rueda, M. R. (2005). The development of effortful control. In U. Mayr, E.
Awh, & S. W. Keele (Eds.), Developing individuality in the human brain: A tribute to
Michael I. Posner (pp. 167-188). Washington, DC: American Psychological Association.
100
Rypma, B., Berger, J. S., Prabhakaran, V., Bly, B. M., Kimberg, D. Y., Biswal, B. B., &
D'Esposito, M. (2006). Neural correlates of cognitive efficiency. Neuroimage, 33(3),
969-979.
Samuelsen, C. L., & Fontanini, A. (2017). Processing of intraoral olfactory and gustatory signals
in the gustatory cortex of awake rats. Journal of Neuroscience, 37(2), 244-257.
Sarinopoulos, I., Grupe, D. W., Mackiewicz, K. L., Herrington, J. D., Lor, M., Steege, E. E., &
Nitschke, J. B. (2010). Uncertainty during anticipation modulates neural responses to
aversion in human insula and amygdala. Cerebral Cortex, 20(4), 929-940.
doi:10.1093/cercor/bhp155
Schatz, J. (1998). Cognitive processing efficiency in schizophrenia: Generalized vs domain
specific deficits. Schizophrenia Research, 30(1), 41-49.
Schiff, H. C., Bouhuis, A. L., Yu, K., Penzo, M. A., Li, H., He, M., & Li, B. (2018). An Insula-
Central Amygdala Circuit for Guiding Tastant-Reinforced Choice Behavior. Journal of
Neuroscience, 38(6), 1418-1429. doi:10.1523/JNEUROSCI.1773-17.2017
Schwab, J., Bialow, M., Clemmons, R., Martin, P., & Holzer, C. (1967). The Beck Depression
Inventory with medical inpatients. Acta Psychiatrica Scandinavica, 43, 255-266.
Seeley, W. W., Menon, V., Schatzberg, A. F., Keller, J., Glover, G. H., Kenna, H., . . . Greicius,
M. D. (2007). Dissociable intrinsic connectivity networks for salience processing and
executive control. Journal of Neuroscience, 27(9), 2349-2356.
Shidara, M., & Richmond, B. J. (2002). Anterior cingulate: single neuronal signals related to
degree of reward expectancy. Science, 296(5573), 1709-1711.
Shomstein, S. (2012). Cognitive functions of the posterior parietal cortex: Top-down and
bottom-up attentional control. Frontiers in integrative neuroscience, 6, 38.
101
Simmons, D. (2008). The use of animal models in studying genetic disease: Transgenesis and
induced mutation. Nature education, 1(1), 70.
Simmons, W. K., Avery, J. A., Barcalow, J. C., Bodurka, J., Drevets, W. C., & Bellgowan, P.
(2013). Keeping the body in mind: Insula functional organization and functional
connectivity integrate interoceptive, exteroceptive, and emotional awareness. Human
brain mapping, 34(11), 2944-2958. doi:10.1002/hbm.22113
Singer, T., Critchley, H. D., & Preuschoff, K. (2009). A common role of insula in feelings,
empathy and uncertainty. Trends in cognitive sciences, 13(8), 334-340.
Singer, T., Seymour, B., O'Doherty, J., Kaube, H., Dolan, R. J., & Frith, C. D. (2004). Empathy
for pain involves the affective but not sensory components of pain. Science, 303(5661),
1157-1162.
Sridharan, D., Levitin, D. J., & Menon, V. (2008). A critical role for the right fronto-insular
cortex in switching between central-executive and default-mode networks. Proceedings
of the National Academy of Sciences, 105(34), 12569-12574.
Sterzer, P., & Kleinschmidt, A. (2010). Anterior insula activations in perceptual paradigms:
Often observed but barely understood. Brain Structure and Function, 214(5-6), 611-622.
doi:10.1007/s00429-010-0252-2
Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of experimental
psychology, 18(6), 643-662.
Swick, D., Ashley, V., & Turken, A. U. (2008). Left inferior frontal gyrus is critical for response
inhibition. BMC Neuroscience, 9, 102. doi:10.1186/1471-2202-9-102
102
Swick, D., & Jovanovic, J. (2002). Anterior cingulate cortex and the Stroop task:
Neuropsychological evidence for topographic specificity. Neuropsychologia, 40(8),
1240-1253.
Sych, Y., Chernysheva, M., Sumanovski, L. T., & Helmchen, F. (2019). High-density multi-fiber
photometry for studying large-scale brain circuit dynamics. Nature Methods, 16, 553-
560. doi:10.1101/422857
Taylor, K. S., Seminowicz, D. A., & Davis, K. D. (2009). Two systems of resting state
connectivity between the insula and cingulate cortex. Human brain mapping, 30(9),
2731-2745. doi:10.1002/hbm.20705
Tomasino, B., Marin, D., Canderan, C., Maieron, M., Skrap, M., & Ida Rumiati, R. (2014).
Neuropsychological Patterns Following Lesions of the Anterior Insula in a Series of
Forty Neurosurgical Patients. AIMS Neuroscience, 1(3), 225-244.
doi:10.3934/Neuroscience.2014.3.225
Trautwein, F. M., Singer, T., & Kanske, P. (2016). Stimulus-Driven Reorienting Impairs
Executive Control of Attention: Evidence for a Common Bottleneck in Anterior Insula.
Cerebral Cortex. doi:10.1093/cercor/bhw225
Tsakiris, M., Hesse, M. D., Boy, C., Haggard, P., & Fink, G. R. (2007). Neural signatures of
body ownership: A sensory network for bodily self-consciousness. Cerebral Cortex,
17(10), 2235-2244. doi:10.1093/cercor/bhl131
Turken, U., & Swick, D. (1999). Response selection in the human anterior cingulate cortex.
Nature neuroscience, 2(10), 920-924.
Uddin, L. Q. (2015). Salience processing and insular cortical function and dysfunction. Nature
reviews neuroscience, 16(1), 55-61. doi:10.1038/nrn3857
103
van de Werd, H., Rajkowska, G., Evers, P., & Uylings, H. B. (2010). Cytoarchitectonic and
chemoarchitectonic characterization of the prefrontal cortical areas in the mouse. Brain
Structure and Function, 214(4), 339-353.
Varjacic, A., Mantini, D., Levenstein, J., Slavkova, E. D., Demeyere, N., & Gillebert, C. R.
(2018). The role of left insula in executive set-switching: Lesion evidence from an acute
stroke cohort. Cortex, 107, 92-101. doi:10.1016/j.cortex.2017.11.009
Vendrell, P., Junqué, C., Pujol, J., Jurado, M. A., Molet, J., & Grafman, J. (1995). The role of
prefrontal regions in the Stroop task. Neuropsychologia, 33(3), 341-352.
Wang, Q., Ng, L., Harris, J. A., Feng, D., Li, Y., Royall, J. J., . . . Zeng, H. (2017). Organization
of the connections between claustrum and cortex in the mouse. Journal of Comparative
Neurology, 525(6), 1317-1346. doi:10.1002/cne.24047
Wang, X., Wu, Q., Egan, L., Gu, X., Liu, P., Gu, H., . . . Gao, Z. (2019). Anterior insular cortex
plays a critical role in interoceptive attention. eLife, 8, e42265.
Weller, J. A., Levin, I. P., Shiv, B., & Bechara, A. (2009). The effects of insula damage on
decision-making for risky gains and losses. Soc Neurosci, 4(4), 347-358.
doi:10.1080/17470910902934400
Wetzels, R., Raaijmakers, J. G., Jakab, E., & Wagenmakers, E.-J. (2009). How to quantify
support for and against the null hypothesis: A flexible WinBUGS implementation of a
default Bayesian t test. Psychonomic bulletin & review, 16(4), 752-760.
White, M. G., Panicker, M., Mu, C., Carter, A. M., Roberts, B. M., Dharmasri, P. A., & Mathur,
B. N. (2018). Anterior Cingulate Cortex Input to the Claustrum Is Required for Top-
Down Action Control. Cell Reports, 22(1), 84-95. doi:10.1016/j.celrep.2017.12.023
104
Wimmer, R. D., Schmitt, L. I., Davidson, T. J., Nakajima, M., Deisseroth, K., & Halassa, M. M.
(2015). Thalamic control of sensory selection in divided attention. Nature, 526(7575),
705-709. doi:10.1038/nature15398
Wu, Q., Chang, C. F., Xi, S., Huang, I. W., Liu, Z., Juan, C. H., . . . Fan, J. (2015). A critical role
of temporoparietal junction in the integration of top‐down and bottom‐up attentional
control. Human brain mapping, 36(11), 4317-4333.
Wu, T., Chen, C., Spagna, A., Wu, X., Mackie, M. A., Russell-Giller, S., . . . Fan, J. (2020). The
functional anatomy of cognitive control: A domain-general brain network for uncertainty
processing. Journal of Comparative Neurology, 528, 1265-1292. doi:10.1002/cne.24804
Wu, T., Dufford, A. J., Egan, L. J., Mackie, M. A., Chen, C., Yuan, C., . . . Fan, J. (2018). Hick-
Hyman Law is Mediated by the Cognitive Control Network in the Brain. Cerebral
Cortex, 28(7), 2267-2282. doi:10.1093/cercor/bhx127
Wu, T., Dufford, A. J., Mackie, M. A., Egan, L. J., & Fan, J. (2016). The Capacity of Cognitive
Control Estimated from a Perceptual Decision Making Task. Scientific Reports, 6, 34025.
doi:10.1038/srep34025
Wu, T., Wang, X., Wu, Q., Spagna, A., Yang, J., Yuan, C., . . . Fan, J. (2019). Anterior insular
cortex is a bottleneck of cognitive control. Neuroimage, 195, 490-504.
doi:10.1016/j.neuroimage.2019.02.042
Xuan, B., Mackie, M. A., Spagna, A., Wu, T., Tian, Y., Hof, P. R., & Fan, J. (2016). The
activation of interactive attentional networks. Neuroimage, 129, 308-319.
doi:10.1016/j.neuroimage.2016.01.017
105
Yamamoto, K., Koyanagi, Y., Koshikawa, N., & Kobayashi, M. (2010). Postsynaptic cell type–
dependent cholinergic regulation of GABAergic synaptic transmission in rat insular
cortex. Journal of neurophysiology, 104(4), 1933-1945.
Yamamoto, T., Yuyama, N., Kato, T., & Kawamura, Y. (1985). Gustatory responses of cortical
neurons in rats. II. Information processing of taste quality. Journal of neurophysiology,
53(6), 1356-1369.
Yoshida, W., & Ishii, S. (2006). Resolution of uncertainty in prefrontal cortex. Neuron, 50(5),
781-789.
Young, J. W., Light, G. A., Marston, H. M., Sharp, R., & Geyer, M. A. (2009). The 5-choice
continuous performance test: Evidence for a translational test of vigilance for mice. PLoS
One, 4(1), e4227. doi:10.1371/journal.pone.0004227
Zaki, J., Davis, J. I., & Ochsner, K. N. (2012). Overlapping activity in anterior insula during
interoception and emotional experience. Neuroimage, 62(1), 493-499.
doi:10.1016/j.neuroimage.2012.05.012
Zhang, S., Ide, J. S., & Li, C.-s. R. (2012). Resting-State Functional Connectivity of the Medial
Superior Frontal Cortex. Cerebral Cortex, 22(1), 99-111. doi:10.1093/cercor/bhr088
Zhang, S., Xu, M., Chang, W. C., Ma, C., Hoang Do, J. P., Jeong, D., . . . Dan, Y. (2016).
Organization of long-range inputs and outputs of frontal cortex for top-down control.
Nature neuroscience, 19(12), 1733-1742. doi:10.1038/nn.4417
Zhang, S., Xu, M., Kamigaki, T., Do, J. P. H., Chang, W.-C., Jenvay, S., . . . Dan, Y. (2014).
Long-range and local circuits for top-down modulation of visual cortex processing.
Science, 345(6197), 660-665.
106
Zhou, H., & Desimone, R. (2011). Feature-based attention in the frontal eye field and area V4
during visual search. Neuron, 70(6), 1205-1217.
Zingg, B., Dong, H. W., Tao, H. W., & Zhang, L. I. (2018). Input-output organization of the
mouse claustrum. Journal of Comparative Neurology, 526(15), 2428-2443.
doi:10.1002/cne.24502
107