The 30th Anniversary International Australian Learning Group Conference 2019

24th – 26th July 2019

Peppers Blue on Blue Resort Magnetic Island QLD

Supported by:

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Programme Overview Wednesday, Thursday, Friday, 24 July 2019 25 July 2019 26 July 2019

8:50 Welcome to Maggie 9:00 Peter Lovibond 9:00 Dominic Tran 9:00 Justin Harris 9:20 Justine Greenaway 9:20 Gonzalo Urcelay 9:20 Justine Fam 9:40 Hilary Don 9:40 Ian Johnston 9:40 Francesca Wong 10:00 Julie Chow 10:00 Scott Gwinn 10:00 Dana Leidl 10:20 Jessica Lee 10:20 Jean-Richard-dit-Bressel 10:20 Morning tea 10:40 Morning tea 10:40 Morning tea 10:40 Laura Corbit 11:10 Ralph Miller 11:00 J Bertran-Gonzalez 11:10 Mike Le Pelley 11:30 Steven Glautier 11:20 Nathan Marchant 11:30 Poppy Watson 11:50 Robert Honey 11:40 Charlotte Bonardi 11:50 Luca Blumhardt 12:10 Douglas Elliffe 12:00 Ignacio Loy 12:10 Anna Thowart 12:30 Evan Livesey 12:20 Lunch 12:30 Lunch

12:50 Lunch 13:00 Dominic Dwyer 13:20 Simone Rehn 13:20 Ann Meulders

13:50 Anne Macaskill 13:40 Michael Kendig 13:40 Ann-Kathrin Zenzes 14:10 Ottmar Lipp 14:00 Anthony McGregor 14:00 Kirsten Barnes 14:30 Evelina Glogan 14:20 Camilla Luck 14:50 Ian McLaren 14:20 Afternoon Tea 14:40 Anastasia Chalkia 15:10 Caroline Moul Keynote Address by 14:40 15:00 Afternoon Tea David Shanks Posters & Afternoon Keynote Address by 15:30 15:40 End of presentations 15:20 Tea Kate Wassum History of ALG by 16:30 16:20 End of presentations Bob Boakes 17:00 End of presentations

19:00 Welcome BBQ 19:00 Conference Dinner

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Programme Wednesday 24th July

Session 1 Chair: Anna Thowart

8:50 Welcome Address

9:00 Causal structures in human inhibitory learning. Peter F. Lovibond & Jessica C. Lee University of New South Wales

9:20 Blocking and cognitive reflection in causal learning. Justine Greenaway & Evan J. Livesey University of Sydney

9:40 Frequency effects in causal learning. Hilary J. Don & Darrell A. Worthy Texas A&M University

10:00 Finding the base-rate: Effects of trial sequencing on the strength of illusory causation. Julie Y.L. Chow & Evan J. Livesey University of Sydney

10:20 Modeling individual differences and multiple rules in generalisation. Jessica Lee1, Peter Lovibond1, Brett Hayes1, & Stephan Lewandowsky2 1University of New South Wales, 2University of Bristol

Morning Tea 10:40 – 11:10

Session 2 Chair: Justin Harris

11:10 ACQUISITION of Associative Memory is Driven by TOTAL PREDICTIVE Error-Reduction: No!, No!, and No! Ralph R. Miller & Cody W. Polack State University of New York at Binghamton

11:30 Non-local influences on associative learning: new data and further model evaluation. Steven Glautier Southampton University

11:50 HeiDI: How excitation and inhibition determine ideo-motion Robert C. Honey, Dominic M. Dwyer, & Adela F. Iliescu Cardiff University

12:10 Testing behavioural data for trend using rank permutation. Douglas Elliffe & Martin Elliffe The University of Auckland

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

12:30 Learning to predict neural stimulation: Effects of self-initiation and temporal cuing on motor- evoked potentials. Evan J. Livesey, Dominic M.D. Tran, & Nicolas A. McNair University of Sydney

Lunch 12:50 – 13:50

Session 3 Chair: Gonzalo Urcelay

13:50 Delay discounting and temporal bisection: When people are less willing to wait does time subjectively drag? Anne C. Macaskill, Kate Witt, & Maree J. Hunt Victoria University of Wellington

14:10 Evaluative conditioning affects subsequent fear learning. Ottmar V. Lipp, Camilla C. Luck, & Alana Muir Curtin University

14:30 Generalization of operantly acquired avoidance to novel but similar movements using a robotic arm reaching paradigm. Eveliina Glogan1, 2, Irene Gatzounis1, 2, & Ann Meulders1, 2 1Experimental Health , Maastricht University; 2Research Group Health Psychology, KU Leuven

14:50 Using tDCS to Modulate Perceptual Learning and Face Recognition. Ciro Civile, Rosamund McLaren, & Ian McLaren University of Exeter

15:10 The roots of all evil: can a bias in attention and associative learning explain psychopathy? Caroline Moul1 & Tom Beesley2 1University of Sydney, 2University of Lancaster

15:30 Poster Session and afternoon tea

30th Anniversary Special Talk Chair: Anne Swinbourne

16:30 A history of 30 years of the ALG Bob Boakes University of Sydney

Welcome BBQ Dinner 19:00

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Thursday 25th July

Session 1 Chair: Ian McLaren

9:00 Examining controlled versus involuntary motor preparation with the Perruchet effect and TMS. Dominic M.D. Tran, Justin A. Harris, Irina M. Harris, & Evan J. Livesey University of Sydney

9:20 Delayed consequences promote habits in rats and humans. Gonzalo P. Urcelay1, Selina Chadha1, Sietse Jonkman2, Omar D. Perez3, & Anthony Dickinson4 1University of Leicester, 2The Boston Consulting Group, 3California Institute of Technology, 4University of Cambridge

9:40 Simultaneous acquisition of allocentric and egocentric reference frames in human spatial learning. Ian Johnston, Blake Segula, Kate Thompson University of Sydney

10:00 Electrophysiological investigation of learned attentional biases. O. Scott Gwinn1, Reneshree Govender1, Salvatore Russo2, Irina Baetu2, Mike Nicholls1 & Oren Griffiths1 1Flinders University, 2University of Adelaide

Morning Tea 10:20 – 10:40

Session 2 Chair: Kate Wassum

10:40 The role of noradrenaline in updating predictions. Laura Corbit The University of Toronto

11:00 Interneuron-like role of D2-projection systems in the functional parcellation of striatal plasticity during goal-directed learning. J Bertran-Gonzalez, Bernard W Balleine, & Miriam Matamales Decision Neuroscience Laboratory, School of Psychology, University of New South Wales

11:20 Reward seeking under motivational conflict: behavioral and neuronal mechanisms. Nathan Marchant1 Allison McDonald1, Isis Alonso1, Dustin Schetters1, Sybren de Kloet2, & Taco de Vries1 1 Department of Anatomy & Neurosciences, VUmc, Amsterdam University Medical Center, 2 Department of Integrative Neurophysiology, Center for Neurogenomics and Cognitive Research, VU University

11:40 Recognition memory impairments in APPswe/Ps1dE9 mice: relevance for early diagnosis of Alzheimer's disease. Charlotte Bonardi1, Jasper Robinson1, Marie-Christine Pardon2, & Paul Armstrong3 1School of Psychology, University of Nottingham, 2Faculty of Medicine & Health Sciences, University of Nottingham, 3School of Biomedical Sciences, University of Leeds

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

12:00 Contextual effects by circadian cues in conditioned tentacle lowering in snails Cornu aspersum: Renewal and Latent Inhibition. Judit Muñiz-Moreno, Manuel Rivero, & Ignacio Loy Departamento de Psicología, Universidad de Oviedo

Lunch 12:20 – 13:00

Session 3 Chair: Robert Honey

13:00 Consumption analysis as a welfare measure: Examples from mouse handling and rat intraperitoneal injections. Dominic Dwyer Cardiff University

13:20 Prior exposure to sucrose makes saccharin less acceptable: Only incentive contrast or something more? Simone Rehn1, Robert A. Boakes1 & Kieron B. Rooney2 1 School of Psychology and 2Faculty of Health Sciences, University of Sydney

13:40 Does a cookie a day lead memory astray? Identifying how different schedules of access to unhealthy diets affect cognition in rats. Michael D. Kendig1, R. Fred Westbrook2, & Margaret J. Morris1 1School of Medical Sciences and 2School of Psychology, University of New South Wales

14:00 Spatial stability and cue type do not influence blocking in spatial learning. Anthony McGregor1, Adina Lew2, & Matthew G. Buckley1,3 1Durham University, 2Lancaster University, 3De Montfort University

Afternoon Tea 14:20

International Keynote Address Chair: Peter Lovibond

14:40 Testing your memory: The many consequences of retrieval on long-term learning and retention. David Shanks University College London

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Friday 26th July

Session 1 Chair: Laura Corbit

9:00 Patterns of reinforced and non-reinforced trials during conditioning affect subsequent extinction. Justin Harris & Jonas C. K. Chan University of Sydney

9:20 The perirhinal cortex and basolateral amygdala have different roles in the expression of sensory preconditioned fear. Justine Fam, Vincent Laurent, R. Fred Westbrook, & Nathan M. Holmes University of New South Wales

9:40 The perirhinal cortex encodes, stores and retrieves associations that form in a sensory preconditioning procedure. Francesca Wong, R. Fred Westbrook1 & Nathan M. Holmes1 University of New South Wales

10:00 A “synaptic tag and capture”-like mechanism (probably) does not underlie the consolidation of second-order fear. Dana M. Leidl1, Belinda P. P. Lay2, Frederick Westbrook1, & Nathan M. Holmes1 1University of New South Wales, 2Center for Studies in Behavioural Neurobiology, Department of Psychology Concordia University

10:20 Phasic activity of basolateral amygdala neurons during concurrent Pavlovian and instrumental aversive learning Philip Jean-Richard-dit-Bressel & Gavan P. McNally University of New South Wales

Morning Tea 10:40 – 11:10

Session 2 Chair: David Shanks

11:10 Measuring habit formation through goal-directed response switching. Mike E. Le Pelley1, David Luque2,3, Poppy Watson1, Sara I. Molinero3, & Francisco J. López3 1 University of New South Wales, 2 Universidad Autónoma de Madrid, 3 Universidad de Málaga

11:30 The persistence of value-modulated attentional capture. Poppy Watson, Daniel Pearson, Jenny Le, & Mike E. Le Pelley University of New South Wales

11:50 Investigating the behavioural mechanisms behind value-mediated oculomotor capture. Luca Blumhardt, Evan J. Livesey, & Justin A. Harris University of Sydney

12:10 Effects of previous outcome predictability and incentive value on subsequent learning. Anna Thorwart University Marburg

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Lunch 12:30 – 13:20

Session 3 Chair: Anne Macaskill

13:20 Generalization of pain-related avoidance behavior Anne Meulders KU Leuven & Maastricht University

13:40 Thought conditioning: Inducing and reducing thoughts about the aversive outcome in a fear conditioning procedure. Ann-Kathrin Zenses1, Frank Baeyens1, Tom Beckers1, Yannick Boddez2,1 1Center for the Psychology of Learning and Experimental Psychopathology, KU Leuven; 2Department of Clinical Psychology and Experimental Psychopathology, University of Groningen

14:00 Context Conditioning of GVS-induced nausea in Virtual Reality Environments. Kirsten Barnes, Robert A. Boakes, & Ben Colagiuri University of Sydney

14:20 Reinstatement with Novel Unconditional Stimuli during Evaluative Conditioning – Relapse depends on the Modality of the Reinstatement Unconditional Stimulus Camilla C. Luck, Rachel R. Patterson, & Ottmar V. Lipp Curtin University

14:40 Findings from a highly-powered, direct replication attempt of the reactivation-extinction effect in humans. Anastasia Chalkia, Natalie Schroyens, Ann-Kathrin Zenses, Lukas Van Oudenhove, & Tom Beckers KU Leuven

Afternoon Tea 15:00 – 15:20

International Keynote Address Chair: Peter Lovibond

15:20 Cortical-amygdala circuitry in reward learning and decision making Kate Wassum University of California, Los Angeles

Conference Dinner 19:00

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019, Magnetic Island QLD

International Keynote Addresses

Professor David Shanks, University College London

“Testing your memory: The many consequences of retrieval on long-term learning and retention”.

Memory tests do much more than simply permit an assessment of knowledge, they are learning events in their own right. In the classic testing effect, retention of tested material is enhanced. Here I describe evidence that testing can have another powerful effect, namely to potentiate subsequent learning. The empirical evidence and putative mechanisms underlying this forward testing effect are described. I report experiments showing that the effect generalizes to a broad range of materials (e.g., vocabulary, text, concept learning), shows transfer across domains (a test on one form of material enhances learning of a different form of material) and test types (one type of test can facilitate performance on a different type of test), and is observed with a range of participant populations. Further experiments reveal that tests taken prior to learning can enhance acquisition even when they inevitably yield errorful responses. Strikingly, people are often unaware of the beneficial effects of tests and as a result allocate their study effort irrationally. Deepening understanding of the many consequences of retrieval is beginning to have a significant impact on optimizing learning and teaching in educational settings.

Associate Professor Kate Wassum, University of California, Los Angeles

“Cortical-amygdala circuitry in reward learning and decision making”.

To make adaptive decisions we must cast ourselves into the future and consider the outcomes of our potential choices. This prospective consideration is informed by our memories. I will discuss our lab’s recent work investigating the neural circuits responsible for encoding, updating, and retrieving reward memories for use in the considerations underlying decision making. We have taken a multifaceted approach to these investigations, combining recording, modern circuit dissection, and behavioral tools. Our results indicate that the and basolateral amygdala work in a reciprocal circuit to mediate in these functions. The cognitive symptoms underlying many psychiatric disorders result from a failure to appropriately learn about and/or anticipate potential future events, making these basic science data relevant to the understanding and potential treatment of mental illness.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Talk Abstracts

Wednesday 24th July

Causal structures in human inhibitory learning

Peter F Lovibond and Jessica C. Lee UNSW

Animal research suggests that intermixed A+ and AB- trials endow B with the ability to inhibit other excitors in a summation test, so long as A and B are presented in a simultaneous compound (conditioned inhibition). By contrast, intermixed A+ and B->A- trials, where B precedes A in a serial compound, endow B with the ability to modulate responding to A but not other excitors (negative occasion-setting). In human causal learning, transfer in a summation test is generally weak even after simultaneous compound training, suggesting that humans might be more likely to adopt a modulatory causal structure. Our recent work with generalisation has shown that group-level results can arise from the combination of qualitatively different patterns in distinct subgroups, identifiable by their self-reported rules. In the present research, we similarly investigated whether there are subgroups who report distinct causal structures after conditioned inhibition training and who show correspondingly different patterns of transfer. We addressed this question through assessment of self-reported causal structure after training with a conditioned inhibition design, and then by comparing the degree of transfer in each subgroup in a summation test. We also tested whether we could manipulate the causal structures learned by participants, and their transfer performance, by pre-experimental verbal hints. The implications of the results for the mechanisms of causal and preventive learning will be discussed.

Blocking and cognitive reflection in causal learning

Justine Greenaway1, Jessica Lee2, Thomas Whitford2 & Evan Livesey1 1. University of Sydney; 2. University of New South Wales

Theories of causal learning have described the blocking effect alternately as the result of associative learning, or as the outcome of a more deliberative deductive reasoning process. Consistent with the inferential account, the blocking effect is stronger when participants are encouraged to assume that causes have an additive effect on the outcome, an assumption necessary to deduce that the blocked cue is not causal. These accounts of blocking are not necessarily mutually exclusive however; we may rely on associative memory retrieval to make causal judgements under some circumstances and engage in deliberative inferential reasoning under others. In fact, blocking often persists even when participants hold nonadditive assumptions that should eliminate deductive inference. We examined the relationship between the blocking effect and performance on reasoning and cognitive reflection tasks. By manipulating the assumptions that permit inferential blocking we found that the extent to which participants express blocking depends not only what kind of inference is encouraged, but also on participants’ propensity to engage in critical reflection. Specifically there was a negative

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD relationship between nonadditive training and cognitive reflection, and this relationship was in the opposite direction for those given additivity training. Low scores on the cognitive reflection task were associated with a reliable blocking effect that was less sensitive to manipulations of the assumptions important for inferential blocking. This suggests that causal judgements may reflect both inferential and associative processes, but that blocking in causal learning is often the result of an intuitive or uncritical mode of thinking.

Frequency effects in causal learning

Hilary J. Don & Darrell A. Worthy Texas A&M University

Recent work has provided evidence that people learn more about the frequency of rewarding outcomes than the probability of receiving reward. In this study, we examined how causal judgments are affected by the frequency versus probability of an outcome. Participants were asked to assume the role of a doctor trialing new drug treatments for a disease. On each trial, they were asked to choose between two drug options to administer to a new patient, and observe whether or not the patient’s health improved. During training in the critical group, participants selected between pairs of options that had outcome probabilities of .60 (A) vs .40 (B), or .80 (C) vs .20 (D). Importantly, there were twice as many AB trials than CD trials, such that option A was associated with greater cumulative instances of improved health, while option C had a higher probability of improved health. In a test phase, participants chose between several novel combinations of options. We examined choices on AC trials, where choice of option A indicates a preference for the option providing a frequently occurring outcome, and choice of option C indicates a preference for the option providing higher probability of the outcome. Participants were then asked to rate the efficacy of each drug. Results and implications for causal learning will be discussed.

Finding the base-rate: Effects of trial sequencing on the strength of illusory causation

Julie YL Chow, Jenny Yang, and Evan J Livesey University of Sydney

People are generally good at judging the causal relationship between a putative cause and a probabilistic outcome. For example, causal judgments about the efficacy of a drug treatment (i.e. the cue) usually reflect the extent to which the administration of the treatment cue elevates the probability of recovery (i.e. the outcome) above an observed base-rate. However, when there is no real causal relationship between a cue and an outcome, people display a tendency towards illusory causation—that is, to falsely regard the cue as causing the outcome. Experimental procedures that reliably produce these illusory effects rely on certain properties that limit the individual's ability to calculate and reason appropriately about the base-rate of the outcome. We tested whether one of these properties—the sequencing of trial types, in particular the length of consecutive cue-absent trials (required to estimate the base-rate)—affected the strength of illusory causation. Using a zero contingency task with an ineffective treatment cue and a fictitious medical outcome, we found that

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD clustering of trial types had a substantial impact on illusory causation even when the event frequencies were held constant. Implications for theories of contingency learning and false belief will be discussed.

Modeling individual differences and multiple rules in generalisation

Jessica Lee1, Peter Lovibond1, Brett Hayes1, and Stephan Lewandowsky2 1. University of New South Wales; 2. University of Bristol

Generalisation involves extrapolating from the known to the unknown, and is thus inherently uncertain. This uncertainty often leads to large individual differences in generalisation gradients after training with the same stimuli. For example, after differential training with a CS+ and CS- that differ on a physical dimension (e.g., colour), some participants generalise on the basis of physical similarity to the CS+ (a similarity generalisation rule), while others generalise on the basis of the relationship between the CS+ and CS- (e.g. “CS+ was greener than CS-”, a relational generalisation rule). However, there has been little research on whether it is possible for a given individual to generalise according to multiple generalisation rules. In this talk, I will present a new model linking explicit beliefs in multiple generalisation rules to behaviour. By implementing the model in a hierarchical Bayesian framework, we show how generalisation gradients can be decomposed into multiple underlying components that differ between individuals. Implications for generalisation phenomena such as peak shift will be discussed.

ACQUISITION of Associative Memory is Driven by TOTAL PREDICTIVE Error-Reduction: No!, No!, and No!

Ralph R Miller and Cody W Polack State University of New York at Binghamton

Three dogmas in one sentence. A series of phenomena will be reviewed that indicate: (a) Error between outcome expectation and occurrence modulates subsequent retrieval of associative memories as much or more than it contributes to acquisition of new memories; (b) Reduction in Local (as opposed to Total) error between outcome and expectation suffices to account for negatively accelerated acquisition curves, and cue competition is due largely to retrieval failure as opposed to acquisition failure; and (c) the ‘errors’ that modulate acquisition and retrieval need not be Predictive (i.e., forward). Phenomena reviewed will include the testing effect, Rescorla’s compound testing experiments demonstrating unequal changes in cues conditioned in compound, recovery from cue competition, and backward conditioning. Collectively, these phenomena support discarding the dogmas that ‘ACQUISITION of associative memory is driven by TOTAL PREDICTIVE Error-Reduction,’ and replacing them, less dogmatically, with ‘Local discrepancy drives formation of associative memories, and total discrepancy modulates subsequent retrieval of associative memories.’

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Non-local influences on associative learning: new data and further model evaluation

Steven Glautier Southampton University

Previous work (Glautier, 2013) showed that the responses made by humans on trial n in simple associative learning tasks were influenced by events that took place on trial n−1 and a simple extension of the Rescorla-Wagner Model (RWM Rescorla & Wagner, 1972), the Memory Environment Cue Array (MECA) model, was presented to account for those results. In the current work further evidence of non-local influences on responding during associative learning tasks is presented. The Rescorla-Wagner model and the MECA model are evaluated as models for the observed data using qualitative, näive maximum likelihood, and Akaike weight analyses. In two experiments the Akaike weight analyses strongly supported the simpler Rescorla-Wagner model over the MECA model but the qualitive and näive maximum likelihood analyses strongly supported the MECA model model over the simpler Rescorla-Wagner model. In Experiment 2 this apparent conflict was resolved using a generalisation criterion test (Ahn, Busemeyer, Wagenmakers, & Stout, 2008; Busemeyer & Wang, 2000) which gave clear support to the MECA model over the Rescorla-Wagner model. These results demonstrate the superiority of model selection using predictive validity, where possible, over selection using statistical adjustments for model complexity.

HeiDI: How excitation and inhibition determine ideo-motion

Robert C. Honey, Dominic M. Dwyer and Adela F. Iliescu Cardiff University

Associative treatments of how Pavlovian conditioning affect changes in conditioned behavior or ideo- motion are rudimentary. Their simplifying assumption is that there exists an ordinal mapping between associative strengths (Vs) and conditioned behavior in a given experimental preparation. The inadequacy of this simplification is highlighted by recent studies that have taken multiple measures of conditioned behavior. These studies reveal phenotypic differences in how learning is manifest across different animals: With different measures of conditioned behavior providing the basis for drawing opposite conclusions about Vs. Here, we develop a simple integrated computational model of how learning affects performance that simulates these qualitative individual differences in conditioned behavior. The new model, HeiDI, enables a broad range of phenomena to be accommodated, which are either beyond the scope of extant associative models or require additional (learning) processes. It also provides an impetus for new lines of inquiry and generates novel predictions.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Testing behavioural data for trend using rank permutation

Douglas Elliffe and Martin Elliffe The University of Auckland

We make an argument that exact-probability, rank-permutation null-hypothesis statistical tests are particularly suitable for behavioural data, because these tests avoid both the distributional and measurement assumptions that underpin standard parametric NHSTs. We show that the historical barrier to using these tests, the complexity of enumerating permutations to establish the sampling distribution of the test statistic, can readily be overcome with modern computing power and efficient algorithms. We illustrate this with a rank-permutation test for monotonic trend based on that of Kendall (1955) and extend this test to allow for unequal numbers of data points for each subject. We offer a web-based Java program that almost instantly calculates the sampling distribution and critical values of Kendall’s � even with as many as 20 subjects with 20 data points each. We hope this test will be useful to fellow researchers.

Learning to predict neural stimulation: Effects of self-initiation and temporal cuing on motor-evoked potentials.

Evan J. Livesey, Dominic M. D. Tran, & Nicolas A. McNair The University of Sydney

The brain's response to sensory input is modulated by prediction. For instance, sounds that are produced by one's own actions or that are strongly predicted by other environmental cues are accompanied by an attenuated N1 component in auditory evoked potentials and perceived as being less salient. Here we tested whether the neural response to direct stimulation of the brain is attenuated by prediction in a similar way. Transcranial magnetic stimulation (TMS) applied over primary motor cortex is often used to gauge the excitability of the motor system. Motor-evoked potentials (MEPs), elicited by TMS and measured in peripheral muscles, tend to be larger when actions are being prepared and smaller when actions are voluntarily suppressed. We tested whether the magnitude of the MEP was attenuated under circumstances where the TMS pulse can be reliably predicted, even though control of the relevant motor effector was not required. Self-initiation of the TMS pulse and reliable cuing of the TMS pulse were both associated with reduced MEP magnitude. We will discuss how this relates to domain-general predictive learning mechanisms.

Delay discounting and temporal bisection: When people are less willing to wait does time subjectively drag?

Anne C. Macaskill, Kate Witt & Maree J. Hunt Victoria University of Wellington

People often choose a smaller reward now over a larger reward later, and are even more likely to do this when they have to wait for the larger reward. To study this we ask people to make choices like

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD whether to watch five seconds of a funny video now or wait for 15 seconds and then watch 10 seconds of video. Across several experiments in our lab people have become more impulsive during an hour- long experimental session. On the first trial they might happily wait for 10 seconds, but on the 30th trial they are no longer willing to wait. We wondered: is that because time starts to drag as the session wears on? To test this, we had participants complete a temporal bisection task three times during the session. In this task, participants learn that a two-second stimulus is “short” and a four-second stimulus is “long” and are then asked to classify intermediate stimuli (e.g three seconds) as “short” or “long”. If time is subjectively “dragging” a participant will classify a larger number of intermediate stimuli as long. There was no change across the session in participants’ subjective time perception even while willingness to wait decreased. However, we included forced choice trials on which participants were reminded that the two second stimulus was short and the four second long, potentially facilitating a kind of recalibration of time perception. Data collection for an experiment removing these forced choice trials is ongoing and results will be reported.

Evaluative conditioning affects subsequent fear learning

Ottmar V. Lipp, Camilla C. Luck & Alana Muir Curtin University

Evaluative and fear conditioning are two rather different paradigms used to investigate the acquisition of dislikes. However, it is not clear whether the acquisition of negative valence in these procedures reflects on the same or distinct learning mechanisms. The current study used a transfer paradigm to address this question. Three groups of participants (N = 85) were trained in differential fear conditioning comprising acquisition, extinction, and reinstatement test, after completing a picture- picture evaluative conditioning paradigm. Group Congruent was presented with to-be-CS+ negative picture (NP) pairings whereas the to-be-CS- was paired with positive pictures (PP). Group Incongruent was trained with to-be-CS+ PP and to-be-CS- NP pairings, and different CSs were used in evaluative and fear conditioning in group Different. Online measures of CS valence indicated that evaluative conditioning affected valence acquisition during fear conditioning with CS+ being less pleasant than CS- in groups Congruent and Different, but not in group Incongruent. Extinction and reinstatement of negative valence were not affected differentially. Differential electrodermal responses emerged within fewer training trials in groups Congruent and Different than in group Incongruent and there was a trend towards faster extinction in group Incongruent. Reinstatement of electrodermal responses was not affected differentially. The current research indicates that CS valence acquired during evaluative conditioning transfers across conditioning paradigms and will differentially affect the acquisition of fear learning as indexed by subjective evaluations and electrodermal responses. This supports the notion that a single learning mechanisms mediates affective learning in evaluative and fear conditioning.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Generalization of operantly acquired avoidance to novel but similar movements using a robotic arm reaching paradigm

Eveliina Glogan1, 2, Irene Gatzounis1, 2, & Ann Meulders1, 2 1. Maastricht University; 2. KU Leuven

Pain-related avoidance behavior is considered a key contributor to the transition from acute to chronic pain disability. One mechanism through which avoidance may lead to disability is excessive generalization of avoidance to safe movements/activities. This study investigated whether acquired avoidance behavior generalizes to novel but similar movements. In a robotic arm reaching task, participants moved their arm from a starting point to a target via one of three possible movement trajectories. For the Experimental Group, the shortest, easiest trajectory was always paired with a painful electrocutaneous stimulus (T1 = 100% reinforcement/no resistance). The pain could be partly or completely averted by choosing increasingly effortful movements (T2 = 50% reinforcement, moderate resistance/deviation; T3 = 0% reinforcement, strongest resistance/largest deviation). A Yoked Group received the same number of painful stimuli irrespective of their chosen movement trajectory. Dependent variables were self-reported fear of movement-related pain and pain- expectancy, and avoidance behavior –operationalized as the maximal deviation from the shortest trajectory. Generalization was tested to three novel trajectories positioned between the acquisition trajectories (G1-3). Acquired fear of movement-related pain and pain-expectancy generalized to the novel trajectories in the Experimental Group but not in the Yoked Group, and extinguished quickly. In contrast, most participants did not generalize acquired avoidance behavior. These results suggest that avoidance behavior is not directly related to fear- and expectancy-beliefs, and that more deliberate decision-making processes may be involved. However, methodological characteristics may also explain the current findings, and replication is needed to further confirm a possible divergence between these measures.

Using tDCS to Modulate Perceptual Learning and Face Recognition

Ciro Civile, Rosamund McLaren and Ian McLaren University of Exeter

We have provided evidence that perceptual learning can lead to an inversion effect with artificial stimuli (chequerboards) that is analogous to that usually found in faces by adopting an old/new recognition paradigm conventionally used in studies that obtain this effect. Our next step has been to investigate how a particular type of brain stimulation, tDCS, could modulate this effect by reducing the salience modulation that leads to perceptual learning. In recent studies, we have demonstrated that anodal tDCS delivered over the left DLPFC at Fp3 site can eliminate the inversion effect that can otherwise be obtained with chequerboards drawn from a familiar category. Importantly, we have also collected data showing that the same tDCS paradigm can affect the traditional face inversion effect by reducing performance at recognising upright faces. Furthermore, using an active control study (same behavioural task but different tDCS targeted brain area) we confirmed that our anodal tDCS effect is montage-specific. We now have data showing that brief tDCS stimulation over the Fp3 can affect perceptual learning both for chequerboards and for faces, lending more support to the case for

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD perceptual learning being involved in face recognition. In addition, the fact that our tDCS paradigm specifically reduces recognition of upright faces has allowed us to induce a partial analogue of prosopagnosia in healthy participants in the laboratory.

The roots of all evil: can a bias in attention and associative learning explain psychopathy?

Caroline Moul1 and Tom Beesley2 1. University of Sydney; 2. University of Lancaster

Psychopathic personality traits comprise low levels of empathy, a deficient affective response, and diminished feelings of shame and guilt. These features are so at odds with what we consider to be fundamental human characteristics that it is difficult to imagine the cause of psychopathy to be anything but dramatic – childhood trauma and brain damage are commonly cited in the popular media. However, psychopathic traits vary normally within the population and are associated with deficits in performance in cognitive tasks such as response-reversal, the conditioned fear response and emotion-recognition. While these deficits themselves are the focus of much psychopathy research, recent theories argue that each of these deficits could be underpinned by central differences in the allocation of attention and a bias in associative learning. This talk will describe research to test the relationship between psychopathic traits and the allocation of attention and outcome encoding in a basic learning task. The data from a large sample of 244 healthy adults suggests that there are subtle but significant differences in the way in which outcome events are encoded in those with high levels of psychopathic traits. The implications of this finding and the direction of future research will be discussed.

Thursday 25th July

Examining controlled versus involuntary motor preparation with the Perruchet effect and TMS.

Dominic M. D. Tran, Justin A. Harris, Irina M. Harris, & Evan J. Livesey The University of Sydney

The Perruchet effect is often considered to be the best available evidence for independent learning processes in human conditioning. It refers to a dissociation between conscious expectancy of an outcome and the strength of conditioned priming in anticipation of the outcome. However, the standard Perruchet design confounds conditioning history with outcome and response history. Therefore, the dissociation may not be associatively-based, but instead recency-based. Here, we provide a solution to this methodological problem by using transcranial magnetic stimulation (TMS) to probe motor preparation—using motor-evoked potentials (MEPs) to measure conditioned responding. In Experiment 1, participants performed a go/no-go task where they responded to "GO" and withheld responses to "STOP", which was signalled by a fixation cross serving as a preparatory stimulus. In the intertrial interval, participants rated their expectancy that the next trial would be “GO” vs. “STOP”. Critically, participants received TMS on every trial, either before (Before-Cue) or during the fixation cross (During-Cue). In Experiment 2, we adopted the During-Cue TMS design from

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Experiment 1 with a go/no-go task signalled by a single warning cue (replication) or by two predictive warning cues, one that uniquely preceded “GO” trials and the other uniquely preceded “STOP” trials. Results will be discussed in relation to how motor preparation is modulated in the presence and absence of the preparatory stimulus, and how this effect changes when the preparatory stimulus is partially versus fully predictive of the requirement to respond.

Delayed consequences promote habits in rats and humans.

Gonzalo P. Urcelay1, Selina Chadha1, Sietse Jonkman2, Omar D. Perez3, & Anthony Dickinson4 1. University of Leicester; 2. The Boston Consulting Group; 3. California Institute of Technology; 4. University of Cambridge

Instrumental behaviour can be goal-directed or habitual, depending variables such as amount of training, schedules of reinforcement, and the use of choice procedures. We will present data from rodents and humans suggesting that another factor, a delay between action and outcome, can determine habitual control over instrumental behaviour. We trained rats to lever-press for pellets presented either immediately or 20 sec later (delayed). Following satiety-specific outcome devaluation, in an extinction test we observed goal-directed control only in the immediate condition (i.e., habits with delayed rewards). In humans, we developed a novel task in which participants purchased shares of different companies, with feedback following each action presented either immediately after a purchase, or 5 sec later (delayed). Following training, we revalued the companies by informing participants that one company crashed, whilst the other company was doing better. This was followed by a test on extinction. We observed significant revaluation when training was conducted with immediate outcomes, but not when these were delayed. Because human actions are most often than not followed by delayed consequences, these results have important implications for human cognition.

Simultaneous acquisition of allocentric and egocentric reference frames in human spatial learning.

Ian Johnston, Blake Segula, & Kate Thompson The University of Sydney

People and other animals solve the universal problem of navigating through space with several types of learning strategies. The two main classes of strategies involve representing space relative to the agent (egocentric) or to features of the environment (allocentric). The aim of this project was to assess how people learn to use these different reference frames if both are available within the same task simultaneously. For this, novel mazes were created using virtual reality technology. The mazes were based on the Morris water maze, and participants could solve it simultaneously using allocentric and/or sequential response-based navigation strategies. The experimental results showed participants could consistently navigate allocentrically after little training, while egocentric strategies were acquired more slowly. Conflict tests showed a strong preference for allocentric over egocentric strategies different mazes even though there was strong evidence that participants had simultaneously learned to use both strategies.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Electrophysiological investigation of learned attentional biases.

O. Scott Gwinn1, Reneshree Govender1, Salvatore Russo2, Irina Baetu2, Mike Nicholls1 & Oren Griffiths1 1. Flinders University; 2. University of Adelaide

Measurement of the influence of associative learning on covert attention, or more generally on pre- attentive processes, can pose difficulties for many behavioural tasks. This talk discusses electrophysiological methods we have used to examine the influence of learned associations on changes in pre-attentive processing (using the mismatch negativity effect), and in both emotion- driven attention and contingency-driven attention (using steady-state visual evoked potentials, or SSVEPs). Of particular interest is whether learned value or learned relevance shapes the deployment of covert attentional processes, and whether EEG methods allow this deployment to be reliably measured trial-by-trial (e.g. for the purposes of constituting an input for computational modelling). In general, our results favour the Macktinosh model, and in particular, more recent value-driven extensions of it (e.g. Le Pelley et al, 2016).

The role of noradrenaline in updating predictions.

Laura Corbit The University of Toronto

Predictive learning allows us and other animals to use information in the environment to anticipate and prepare for biologically significant events and is therefore highly adaptive. Equally important is the ability to update previously learned associations when confronted with new information and environmental contingencies. While substantial research implicates dopamine in initial learning about events, recent research from our lab has shown that updating this learning, notably when reward is omitted, involves noradrenaline. I will present experiments examining the broader role of NA in predictive learning including learning about unexpected reward omission and circumstances where reward is either greater or less than expected.

Interneuron-like role of D2-projection systems in the functional parcellation of striatal plasticity during goal-directed learning

J Bertran-Gonzalez, Bernard W Balleine, & Miriam Matamales UNSW

Nimbly acquiring and extinguishing goal-directed actions is critical to survival in changing environments. While such adaptations seem to depend on the , the way its different subcircuits interact with one another to introduce changes in goal-directed action remains unclear. As the primary integrative input to the basal ganglia, the is mainly composed of two intermingled and equally large projection systems based on the expression of two types of dopamine receptors (i.e. D1 and D2) as well as their output targets. Here, we combined instrumental conditioning with high-throughput reconstruction of nuclear signalling activity in large ensembles of

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD neurons to investigate the function D1- and D2- spiny projection neurons (SPNs) at different stages of goal-directed learning. Our results revealed that D2-SPNs can occupy and suppress D1-SPN territories in a learning-dependent manner, a modulation that appears critical for promoting extinction of previously acquired instrumental behaviours. Consistently, by using new viral approaches based on transneuronal labelling of thousands of neurons, we found that D2-SPNs present a remarkable trans- connectivity towards D1-SPNs, and that former pharmacological stimulation of D2-SPNs can completely prevent the subsequent stimulation of D1-SPNs. Genetic ablation of D2-SPNs in defined striatal territories did not affect acquisition of instrumental behaviour, although significantly delayed its extinction. Our findings suggest that, in a similar way to interneuronal systems, D2-SPNs may exert the function of clearing out obsolete D1-SPN plasticity in the striatum, an extensive parcellation process that allows for efficient shaping of neuronal circuits during goal-directed learning.

Reward seeking under motivational conflict: behavioral and neuronal mechanisms.

Nathan Marchant1, Allison McDonald1, Isis Alonso1, Dustin Schetters1, Sybren de Kloet2, Taco de Vries1 1. Amsterdam University Medical Center; 2. Center for Neurogenomics and Cognitive Research, VU University

The decision to seek reward despite risk of danger is a ubiquitous dilemma that organisms face for survival. When confronted with a decision between competing outcomes, a state of motivational conflict arises, and survival depends on selection of the most effective response (either approach or retreat). Ineffective decision making in the face of conflict greatly increases the risk of psychological pathologies, such as substance use disorder. Here we designed a within-session, trial based task to study motivational conflict. In addition, we used fiber photometry to record activity in dmPFCàNAc Core neurons. Long-Evans rats were first injected with retroAAV encoding Cre in NAc Core, and into dmPFC we injected AAV encoding Cre-dependent GCaMP6m, the optic fiber was implanted above dmPFC. We then trained the rats to lever press for a food pellet reward in the presence of the reward discriminative stimulus (rewDS). We then introduced the punishment discriminative stimulus (punDS), during these trials lever press resulted in electric footshock (punishment) without reward. For conflict trials, both cues were presented, and the probability of reward or punishment was 50%. We recorded calcium signal in dmPFCàNAc Core neurons over multiple sessions. We find that lever presses during conflict trials is significantly higher than punishment. Increased shock intensity reduced responding both during punishment and conflict without changing lever presses during reward trials. Preliminary photometry data will be presented identifying the encoding patterns of dmPFCàNAc Core neurons during rewDS and punDS tirals, and compare them to conflict trials.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Recognition memory impairments in APPswe/Ps1dE9 mice: relevance for early diagnosis of Alzheimer's disease

Charlotte Bonardi1, Jasper Robinson1, Marie-Christine Pardon2, & Paul Armstrong3, 1. University of Nottingham; 2. University of Nottingham; 3. University of Leeds

Recognition memory impairment is a key symptom of Alzheimer's disease (AD), and could provide a basis for early diagnosis. This possibility was examined using a mouse model of AD, the APPswe/PS1dE9 mouse. Performance on the spontaneous object recognition (SOR) task was examined in 5-month old APPswe/PS1dE9 mice and wild-type littermate controls - an age at which the neuropathology characterising AD is just beginning to develop. The task variants employed were informed by an analysis of recognition memory performance in terms of Wagner's SOP theory (Robinson & Bonardi, 2015), according to which recognition memory relies on two underlying processes - self-generated priming and retrieval-generated priming. In two experiments we failed to observe a deficit in the standard SOR task, or in relative recency; however abnormalities in an object- in-place task were observed in the transgenic animals. The practical and theoretical implications of these results will be discussed.

Contextual effects by circadian cues in conditioned tentacle lowering in snails Cornu aspersum: Renewal and Latent Inhibition

Judit Muñiz-Moreno, Manuel Rivero and Ignacio Loy Universidad de Oviedo

Previous experiments using tentacle lowering conditioning in terrestrial snails Cornu aspersum have shown extinction and a recovery of the conditioned response (CR) as a consequence of both inserting a delay between the extinction and test (Spontaneous Recovery) and re-exposing to the unconditioned stimulus after extinction (Reinstatement). To complete this research on extinction phenomena, Experiment 1 explored the recovery of the CR due to a change in context (Renewal effect). Thus, conditioning and extinction of an odour CS took place in the presence of different circadian contextual cues (hour of the day and presence of light). The results showed that, returning to the original context of training, defined by circadian cues, produced a recovery of the CR compared to suitable control groups. In Experiment 2 context specificity of Latent Inhibition is evaluated using circadian cues. Subjects received non-reinforced presentations of an odour CS in the same or in a different phase of circadian cycle that the posterior conditioning with that odour. Parallels and discrepancies between the non-reinforced presentations of the CS before (Latent Inhibition) or after (Renewal effect) are discussed. This research was supported by Ministerio de Economía y Competitividad, PSI2015-66974-P.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Consumption analysis as a welfare measure: Examples from mouse handling and rat intraperitoneal injections.

Dominic Dwyer Cardiff University

Laboratory animal welfare is manifestly important it its own right, and is also of direct scientific importance given the potential for adverse welfare to impact on test results. However, welfare assessment typically relies on simple observation and/or monitoring of basic physiological data (e.g. bodyweight). This may miss subtle changes in general welfare and limits quantification of welfare deficits. The amount and nature of consumption of palatable fluids in rodents is determined by both general state (e.g. stress-based depression models display lower consumption and reduced hedonic reactions) and specific experience (e.g. pairing solutions with nausea reduces subsequent consumption and elicits aversive hedonic reactions). Thus consumption analysis offers a potential means for making sensitive assessments of both general welfare and of specific experimental or husbandry manipulations. Here, I provide two examples of this potential: examining the effects of tunnel-handling vs tail-handling in mice, and the impact of vehicle volume for intraperitoneal injections in rats. Tail-handling in mice produced an analogue of anhedonia similar to that seen in explicit depression models – reinforcing current guidance to avoid this handling method. Physiologically inert vehicle injections reduced consumption of a paired flavour without materially affecting hedonic reactions, but there was no suggestion that the effect was reduced by limiting the volume below current recommended maximum levels (20ml/kg). These adverse effects were not apparent from unaided observation or bodyweight changes. I will discuss the implications of these examples for consumption analysis as a flexible welfare assessment tool.

Prior exposure to sucrose makes saccharin less acceptable: Only incentive contrast or something more?

Simone Rehn1, Robert A. Boakes1 and Kieron B. Rooney2 1. School of Psychology and 2. Faculty of Health Sciences, The University of Sydney

In an earlier study intakes of saccharin solutions were unexpectedly low. To discover why this happened, we ran experiments containing three stages: Stage 1, pre-exposure to sucrose versus water for a control group; Stage 2, access to saccharin for 12 days; and Stage 3, test saccharin preferences. In Experiment 1 female rats given access in Stage 1 to either 10% sucrose or 10% glucose, when switched in Stage 2 to saccharin showed both an immediate incentive contrast effect and reduced saccharin acceptance throughout this stage. In Stage 3 the Water group displayed a preference for the saccharin solution over 2% sucrose, but this was not found in the other two groups. Experiments 2 and 3 partially replicated these results with male rats. We are now testing whether incentive contrast depends on the palatability of the initial solution, whereas any maintained reduction in saccharin acceptance is due to an associative process involving ‘missing calories’.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Does a cookie a day lead memory astray? Identifying how different schedules of access to unhealthy diets affect cognition in rats

Michael D. Kendig1, R. Fred Westbrook2, & Margaret J. Morris1 1. School of Medical Sciences and 2. School of Psychology, UNSW Sydney

Diets high in saturated fat and sugar are associated with poorer hippocampal-dependent memory in cross-sectional studies in humans and experimental studies in rodents. Whereas most rodent research has studied cognitive function after continuous access to such diets, less is known about the effects of intermittent consumption, which may better reflect dietary habits for much of the population. Here we compared continuous and intermittent access to a western-style ‘cafeteria’ diet on place and object recognition memory in adult male rats. The duration of exposure to CAF was matched across groups to ensure that differences were attributable to the pattern of access. In Experiment 1, exposure to CAF for 5 or 7 consecutive days per week (23-25 days CAF) impaired place recognition memory and increased fat mass relative to controls fed only chow. In contrast, place recognition memory was unaffected in rats given 3 consecutive days of CAF followed by 4 days of chow. To identify whether the CAF or chow portion of the cycle contributed to intact memory in this group, Experiment 2 compared rats fed CAF for 16 days with rats alternating between 4 days of chow and either 2, 4 or 8 days of CAF. Place recognition memory was poorer in rats given CAF in 16-day than 2-day blocks, indicating that the pattern of access to an unhealthy diet differentially affects cognition even when total exposure to that diet is controlled. Results from a third experiment, which controls time on CAF while varying time on chow, will be presented.

Spatial stability and cue type do not influence blocking in spatial learning

Anthony McGregor1, Adina Lew2, & Matthew G. Buckley1,3 1. Durham University; 2. Lancaster University; 3. De Montfort University

Some theories of spatial learning predict that associative rules apply under only limited circumstances. According to Doeller and Burgess (2008) learning based on a boundary is immune to cue competition because boundary information is the basis for the formation of a cognitive map, whilst landmark learning does not involve cognitive mapping. However, cue stability has been claimed to be a prerequisite for the formation of a cognitive map (O’Keefe & Nadel, 1978), meaning that whichever cue type was perceived as stable would enter a cognitive map and thus be immune to cue competition. Unstable cues will be subject to cue competition, regardless of cue type. We manipulated the stability of boundary and landmark cues when learning the locations of hidden goals. One goal location was constant relative to the boundary, and the other constant relative to landmarks. For one group landmark cues were unstable relative to the boundary, and for another the boundary was unstable relative to landmarks. In a second stage all cues remained stable so that both goal locations could be learned with respect to both landmark and boundary information. According to Doeller and Burgess, boundary information should block learning about landmarks regardless of cue stability. According to the cue stability hypothesis, landmarks should block learning about the boundary when the landmarks appear stable relative to the boundary. Regardless of cue type or

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD stability the results showed reciprocal blocking, contrary to both theories of spatial learning. The results were, however, consistent with an associative analysis of spatial learning.

Friday 26th July

Patterns of reinforced and non-reinforced trials during conditioning affect subsequent extinction

Justin Harris and Jonas C. K. Chan The University of Sydney

If a CS or response has been inconsistently reinforced (i.e., only some trials are followed by the US), conditioned responding will take longer to extinguish than if responding had been established by consistent reinforcement. This Partial Reinforcement Extinction Effect (PREE) is one of the best known phenomena in associative learning but defies ready explanation by associative models which assume that a partial reinforcement schedule will produce weaker conditioning that should be less resistant to extinction. One explanation of the PREE is that animals learn to expect that the US occurs on only some trials during partial reinforcement, and therefore they require a larger number of trials to recognise the absence of the US during extinction. According to earlier accounts, during partial reinforcement animals learn that non-reinforced trials are associated with subsequent reinforcement, and therefore the presence of non-reinforced trials during extinction serves to promote generalisation of conditioning to extinction. One of these accounts argues that animals can encode whole sequences (runs) of non-reinforced trials and associate the sequence with subsequent reinforcement. The present experiment tests this idea directly by comparing extinction of Pavlovian conditioning after partial reinforcement schedules that are matched for the overall percent of reinforced trials but which differ in the length of runs of non-reinforced trials that were experienced during conditioning.

The perirhinal cortex and basolateral amygdala have different roles in the expression of sensory preconditioned fear

Justine Fam, Vincent Laurent, R. Fred Westbrook, & Nathan M. Holmes University of New South Wales

In a protocol known as sensory preconditioning, animals (e.g. rats) form an association between a sound and a light in stage 1. Then, in stage 2, the light is paired with shock. When tested subsequently for fear to both the sound and the light, rats show fear (freezing) to the light as well as the sound, even though the sound had not been explicitly paired with shock. Recent work from our laboratory has shown that the perirhinal cortex (PRh) is important for stage 1, while the basolateral amygdala (BLA) is critical for stage 2, hence, the PRh and the BLA play dissociable roles in the acquisition of sensory preconditioned fear. Here, we use chemogenetic techniques to examine the role of these brain regions in the expression of sensory preconditioned fear to the sound and light across multiple tests. Inhibition of the BLA prior to test 1 reduced freezing at test 1, but increased freezing during test

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

2, which was conducted without BLA inactivation. The acute reduction in freezing at test 1 was mediated by the levels of fear, as it was observed during test of the sensory pre-conditioned sound, which elicits low freezing, but not during the test of the directly conditioned light, which elicits high freezing. Accordingly, when fear to the directly-conditioned light was reduced across additional testing, BLA inactivation produced a reduction in freezing. Overall, inhibition of the BLA reduces the acute expression of both sensory pre-conditioned and directly conditioned fear, but prevents the extinction of that fear.

The perirhinal cortex encodes, stores and retrieves associations that form in a sensory preconditioning procedure.

Francesca Wong, Fred Westbrook, & Nathan Holmes University of New South Wales

How does the brain integrate a past sensory memory with a new fear memory so that a stimulus never associated with danger becomes frightening? The present study addressed this question using an animal model, sensory preconditioning in rats. In this model, rats form a sensory sound-light memory in stage 1, and 24 hours later, a light-shock fear memory in stage 2. Following this training, rats show fear when tested with the directly conditioned light and when tested with the preconditioned sound. Here we show that the ability of the sound to elicit fear depends on the “online” integration of sound-light and light-shock memories during stage 2. When activity in the region that consolidated the sensory sound-light memory (the perirhinal cortex, PRh) was inhibited during or after stage 2, test presentations of the conditioned light elicited fear but test presentations of the preconditioned sound did not. Moreover, when the sound-light and light-shock memories were integrated, stimulus information about the light was re-coded so that it ceased to be represented in the PRh: i.e. activity in the PRh was required for retrieval/expression of fear to the preconditioned sound but not fear to the conditioned light. Together, these results show that the elements of a past sensory memory can associate with those of a newly formed fear memory even though they have never been experienced together; and that the neural representations of sensory stimuli can be altered by experience.

A “synaptic tag and capture”-like mechanism (probably) does not underlie the consolidation of second-order fear

Dana M. Leidl1, Belinda P. P. Lay2, R. Fred Westbrook1, & Nathan M. Holmes1 1. University of New South Wales; 2. Concordia University

The present experiments originated in our recent findings (Lay et al, 2018) that consolidation of a second-order fear memory requires neuronal activity, but not de novo protein synthesis, in the basolateral amygdala complex (BLA). They tested the hypothesis that proteins synthesized in the BLA to consolidate first-order conditioned fear are exploited to consolidate second-order conditioned fear, similar to “synaptic tag and capture” (STC) identified in brain slice electrophysiology studies. They did so by reducing the interval between S1-shock pairings (1st order) and S2-S1 parings (2nd

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD order) from 48 hours to a few minutes. We reasoned that, if an STC-like mechanism underlies the consolidation of second-order conditioned fear, then blocking protein synthesis in the BLA during or after first-order conditioning should simultaneously disrupt consolidation of both first- and second- order conditioned fear. We found no evidence to support this hypothesis. Instead we consistently disrupted fear to the first-order S1 but left intact fear to the second-order S2. These results confirm that consolidation of first-order conditioned fear requires de novo protein synthesis in the BLA. They additionally show that consolidation of second-order conditioned fear does not involve an STC-like exploitation of proteins in the BLA.

Phasic activity of basolateral amygdala neurons during concurrent Pavlovian and instrumental aversive learning

Philip Jean-Richard-dit-Bressel and Gavan P. McNally University of New South Wales

Aversive reinforcement, such as delivery of footshock, can have two distinct consequences for learning and behaviour. First, it supports learning about its environmental antecedents to imbue such stimuli with the ability to elicit conditioned responses (Pavlovian fear conditioning). Second, it supports learning about its behavioural antecedents and alters the probability that these behaviours will be emitted again in the future (punishment). We report a novel within-subjects task (“Reciprocal Yoking”) that permits matched, concurrent assessment of these two different forms of learning. We show that Pavlovian and instrumental aversive contingencies and associated stimuli exert contrasting control over reward behaviour. The basolateral amygdala (BLA) is essential to both fear and punishment but whether and how these might be differentially encoded in BLA is unclear. We used fibre photometry (calcium indicator GCaMP6f and implanted optic fibre) to measure activity of BLA principal neurons to events and behaviours in the Reciprocal Yoking task. We observed differential coding of these events according to contingencies, in line with distinct roles of BLA in these processes.

Measuring habit formation through goal-directed response switching

Mike E. Le Pelley1, David Luque2,3, Poppy Watson1, Sara I. Molinero3, & Francisco J. López3 1. UNSW Sydney; 2. Universidad Autónoma de Madrid; 3. Universidad de Málaga

Reward learning theory views habits as stimulus–response links formed through extended reward training. Accordingly, animal research has shown that actions that are initially goal-directed can become habitual after operant overtraining. However, a similar demonstration is absent in human research, which poses a serious problem for translational models of behavior. We propose that response-time (RT) switch cost after operant training can be used as a new, reliable marker for the operation of the habit system in humans. We show that RT switch cost demonstrates the properties that would be expected of a habitual behavior: (1) it increases with overtraining; (2) it increases when rewards are larger, and (3) it increases when time pressure is added to the task, thereby hindering the competing goal-directed system. These results offer a promising new method for studying the operaon of the habit system in humans.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

The persistence of value-modulated attentional capture

Poppy Watson, Daniel Pearson, Jenny Le, & Mike E. Le Pelley UNSW Sydney

Recent research shows that stimuli paired with high-value rewards involuntarily capture eye gaze more often than stimuli paired with low-value rewards, an effect known as value-modulated- attentional capture (VMAC). This distraction by cues predictive of high reward has been likened to sign tracking in animals (the tendency to approach and engage with a stimulus that signals reward) which is reportedly resistant to outcome devaluation. We have recently begun to investigate the sensitivity of attentional capture by reward to outcome devaluation using both instructed devaluation and satiation manipulations. The studies reported here examined the influence of outcome devaluation on attention using the VMAC task. Participants performed a visual search task, searching for a unique diamond shape amongst a set of circles. Participants were also instructed that a coloured circle (the distractor) in the display indicated whether a high or low reward was available for correctly locating the diamond target on that trial. Looking at the distractor, however, caused the omission of the signalled reward. We used eye-tracking to measure the proportion of trials in which participants nevertheless looked at the distractor. We then reduced the value of one of the outcomes and examined whether this led to a corresponding shift in the rate of attentional capture by a cue signalling that outcome. I will discuss ongoing studies and implications for understanding the role of attentional capture by reward in maladaptive reward-seeking behaviour.

Investigating the behavioural mechanisms behind value-mediated oculomotor capture.

Luca Blumhardt, Evan J. Livesey, & Justin A. Harris The University of Sydney

Previous research has shown that reliably associating rewarding or aversive outcomes with stimuli enhances the ability of those stimuli to draw oculomotor movements towards them, independent of any perceptual distinctiveness – an effect coined value-mediated oculomotor capture. Previous research suggests that a Pavlovian account seems to explain most of the variability in these types of oculomotor responses, given that much of the previous research has demonstrated a VMOC effect even when it is not necessary for participants to orient towards reward-associated stimuli in order to acquire reward. However, we believe that previous research has not ruled out the possibility that individuals operantly orient towards these stimuli in order to acquire information that those stimuli signal about potential value. We devised an appropriate procedure to test this by presenting reward- associated stimuli at the same time as rewards, after responses are made, to prevent the possibility that any operant saccading response towards those stimuli is incidentally reinforced with information. Before this procedure can be implemented, however, an effect must be established during testing using a traditional procedure – where reward-signalling stimuli appear in a display before responses are made. This is so that any negative finding found with the former procedure cannot be attributed to the fact that an operant account of the effect is true, but that the effect has simply not persisted into testing. Should this significant effect be found during testing with the traditional procedure, this will allow us to proceed with testing the alternative procedure.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Effects of previous outcome predictability and incentive value on subsequent learning

Anna Thorwart Philipps-University Marburg

Outcome predictability (OP) effect refers to the tendency to learn more towards one outcome that has a history of predictability than towards an outcome with a history of unpredictability, even if both are encountered in new situations. One question is whether the OP effect is a product of effects of unpredictability on the outcome’s value for the learner and thereby its efficacy to drive learning. We therefore hypothesized that increasing the value of the unpredictable outcome - participants received more or less points for correct responses in allergy task - would diminish the subsequent OP effect. After a first study revealed neither an effect of predictability nor value, one group of participants in a second experiment received more points for a correct response in the less predictable outcome category than in the predictable outcome category, thereby gaining overall more points for the first than the latter despite its unpredictability. Other participants received more points for the predictable outcome category. There were no effects of predictability in either group. Instead, both learning in the initial training phase as well as subsequent learning differed between groups, suggesting an overall effect of outcome value on the motivation to learn.

Generalization of pain-related avoidance behaviour

Ann Meulders KU Leuven & Maastricht University

Fear-avoidance models posit that pain-related fear and avoidance contribute to the transition from acute to chronic pain. When avoidance behavior serves to reduce/eliminate genuine bodily threat, it is adaptive. Yet, in the absence of actual danger, avoidance behavior may become maladaptive and lead to functional disability. Using a novel robotic arm reaching task, in which participants freely moved their arm to reach a target location using the HapticMaster (i.e. 3-degree of freedom programmable robotic arm). Three trajectories (T1-3) led to the target location. During acquisition, participants performed arm reaching movements in two contexts: a pain-avoidance context (e.g., black background) and a yoked context (e.g., white background). In the pain-avoidance context, a painful stimulus could be partly or completely prevented by performing the more effortful (in terms of distance and exerted force) trajectories T2 and T3; in the yoked context, the same number of pain stimuli was delivered irrespective of the chosen trajectories. Avoidance behavior was operationalized as the maximal deviation of the shortest trajectory; additionally we measured pain-related fear and pain-expectancy for each trajectory. I will present two experiments building on this paradigm, investigating (1) generalization of avoidance behavior to novel perceptually similar contexts (e.g., shades of grey backgrounds), (2), differences in the generalization of avoidance to novel safe contexts in healthy participants with either high or low levels of subclinical anxiety testing the hypothesis that high anxious participants would avoid more in novel safe context compared to low anxious participants.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Thought conditioning: Inducing and reducing thoughts about the aversive outcome in a fear conditioning procedure

Ann-Kathrin Zenses1, Frank Baeyens1, Tom Beckers1, & Yannick Boddez2,1 1. KU Leuven; 2. University of Groningen

The human fear conditioning procedure is one of the most influential procedures to study anxiety disorders. Outcome variables that are traditionally measured in this procedure include avoidance, physiology (e.g., skin conductance) and verbal responses (e.g., outcome expectancies). However, merely thinking about the aversive outcome is typically not assessed. This is surprising because thinking of an aversive event is clinically relevant (e.g., in the form of intrusive thinking) and of considerable theoretical interest. Here we present two studies that (1) tested thinking of an aversive outcome as an additional outcome variable and (2) compared several interventions to reduce it. We found that thinking of an aversive outcome could be successfully conditioned and that extinction training was less successful in reducing it than counterconditioning. An association splitting intervention (i.e., presenting new additional outcomes) was also successful in reducing thoughts about the initial outcome, but only if the new outcomes were positive stimuli. Including thinking of the aversive outcome as an additional outcome variable can be a valuable extension to the fear conditioning procedure, which, in turn, may not only serve to enhance the understanding of anxiety disorders but also to inform their treatment. Context Conditioning of GVS-induced nausea in Virtual Reality Environments

Kirsten Barnes, Robert A. Boakes, & Ben Colagiuri University of Sydney

Psychological factors have been implicated in the experience of nausea, such as the exacerbation of anticipatory and post-treatment symptoms in emetogenic chemotherapy. One such factor concerns Pavlovian conditioning of the nauseous-response to the context in which emetogenic treatment is administered. Despite evidence for context conditioning in animal models of nausea, a recent experiment involving conditioning with nauseating Galvanic Vestibular Stimulation (GVS) in human participants found perfect generalisation of conditioned nausea across contexts, suggesting a lack of any context-nausea learning. Overshadowing, contextual similarity, and lack of context salience, however, may have accounted for this result. The present experiment controlled for these factors by employing a within-subjects discrimination learning design. Virtual Reality (VR) was utilised to increase context salience and disparity, while the tactile experience of GVS was non-predictive of nausea. Across three sessions (2x acquisition, 1x test) participants were immersed in two virtual environments for 15 min each time. During the acquisition phase, one virtual environment (VE) was paired with nauseating bipolar GVS and the other with non-nauseating monopolar GVS. At test, both VEs were paired with monopolar GVS and the nauseous response (self-report) and postural instability (associated with bipolar stimulation and nausea) compared across contexts. Limited evidence for context conditioning was observed, despite the presence of increased nausea in the VE paired with bipolar GVS during acquisition. However, substantial habituation to bipolar GVS across acquisition trials potentially confounded results.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Reinstatement with Novel Unconditional Stimuli during Evaluative Conditioning – Relapse depends on the Modality of the Reinstatement Unconditional Stimulus

Camilla C. Luck, Rachel R. Patterson, & Ottmar V. Lipp, Curtin University

During evaluative conditioning, a neutral conditional stimulus (CS) acquires positive or negative valence when it is paired with a pleasant or unpleasant unconditional stimulus (US). This valence reduces during extinction and re-emerges at reinstatement when the USs are presented again. We examined whether CS valence would relapse after reinstatement using a novel US. In Experiment 1, the positive group received CSv – pleasant image and CSn – neutral image pairings, while, the negative group received CSv – unpleasant image and CSn – neutral image pairings. Differential CS valence and US expectancy emerged throughout acquisition and reduced during extinction. At reinstatement, the US valence was congruent with the valence of the USs used during conditioning, but participants either received the US images from acquisition or novel US images. Differential US expectancy and CS valence re-emerged after reinstatement, regardless of which images were presented. In Experiment 2, a US from a different modality was used at reinstatement. Conditioning training was performed with pleasant and unpleasant US images as in Experiment 1. At reinstatement, however, a loud scream (negative conditioning group) or a pleasant melody (positive conditioning group) was presented. After reinstatement, relapse of differential US expectancy was observed. For CS valence, however, both CSv and CSn became more pleasant after the melody and more unpleasant after the scream. Together, the results suggest that a novel US will induce relapse of differential US expectancy, but relapse of differential CS valence is only observed when a US from the same modality as during conditioning is used at reinstatement.

Findings from a highly-powered, direct replication attempt of the reactivation-extinction effect in humans

Anastasia Chalkia, Natalie Schroyens, Ann-Kathrin Zenses, Lukas Van Oudenhove, & Tom Beckers KU Leuven

In line with the idea that upon retrieval, consolidated memories may become labile and thus temporarily susceptible to modification, Schiller et al. (2010) demonstrated that memory reactivation shortly before extinction training can prevent the later spontaneous recovery of conditioned fear responding that is observed after regular extinction training. The report by Schiller et al. in Nature was highly influential, inspiring numerous attempts at conceptual replication, with variable success. One decade later, however, no direct replication of the original findings has yet been reported. We set out to conduct a pre-registered, highly-powered, direct replication of Schiller et al. (2010, Experiment 1). Data collection was substantially delayed, largely due to our discovery of considerable misrepresentation of the exclusion criteria in the original Schiller et al. report, which led to the publication of an addendum to Schiller et al. in 2018. We eventually managed to attain our pre- registered sample size, to find that we could not observe any beneficial effect of reactivation- extinction over standard extinction training in preventing recovery of conditioned fear. We will share the story behind our replication attempt, highlighting the importance of pre-registration during a

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD critical period in the scientific community where reproducibility has become an issue of considerable concern, and we will discuss the implications of our findings for the reliability of the reactivation- extinction effect in humans in the original report and in general.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Poster Abstracts

An attempt to replicate the blocking effect on a PIT task

Daniel E. Alarcón & Mario A. Laborda University of Chile

In the blocking effect, learning about a CS-US relationship is hindered when the CS is trained together with a cue that already has a predictive relationship with that US. This effect demonstrates that contiguity is not enough to develop associations and is critical to the most influential associative learning models. However, some researchers have failed to replicate this phenomenon. The experiments reported here were a replication attempt using a blocking procedure within a Pavlovian- to-instrumental transfer (PIT) task, in which the associative strength of a CS is measured over the performance of instrumental actions. In three experiments, a group of rats was trained using a blocking procedure (X->US; AX->US), while another was trained as a control (B->US; AX->US). In addition, all animals were trained to press a lever. In Experiments 1-2 the lever was trained before the blocking procedure, whereas in Experiment 3 it was reversed. In all the experiments X was tested by itself (blocking test) and with the lever available (PIT tests). There were no detectable between-group differences in either of the tests. In the PIT tests, X reduced lever responding in both groups but it had no significant effect on magazine entries. In Experiment 2, however, X numerically elevated magazine entries in the control group relative to the blocking group. These experiments were an attempt to replicate blocking and to measure it with an additional test (PIT), but no evidence of blocking was found. This failure might have been caused by parametric differences with previous studies.

Directed forgetting of emotional memories

Anastasia Chalkia1, Ann-Kathrin Zenses1, Lotte Stemerding2, Yannick Boddez3, Merel Kindt2, Lukas Van Oudenhove1, & Tom Beckers1

1. KU Leuven; 2. University of Amsterdam; 3. University of Groningen

Declarative memory encoding is sensitive to disruption, and subjects can be cued to intentionally forget information before it is stored using a directed forgetting (DF) manipulation. Until now, DF has been applied in declarative memory procedures only. Here, we report on a novel fear conditioning procedure aimed to investigate whether this technique can be translated to the study of emotional memories. In a first study, 32 simple line drawings were displayed one at a time, and half of them were followed by an aversive image (CS+), while the other half were not (CS-). An acoustic cue was presented after half of the CS+ and CS- trials, indicating that those trials were to be forgotten. Memory retention was assessed in subsequent free recall and recognition tasks. In line with predictions, we found strong, significant main effects of instruction (p < .001), with subjects recalling and recognizing fewer of the items that were followed by the forget cue. The effect was stronger for CS+ items, demonstrated by

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

significant cue x instruction interactions on both tests. In a follow-up study, we slightly modified our procedure by presenting an aversive image following CS+ items and a neutral image following CS- items. We retained our significant main effects of instruction on both tests of memory retention, however, the forgetting effect was stronger for the CS+ items on the recognition task only. Altogether, our findings provide preliminary evidence for the presence of a DF effect in emotional memory encoding.

The comparison process in rat perceptual learning

Isabel de Brugada, Jesús Sánchez, Ana González & Beatriz Juan Universidad de Granada

Perceptual Learning refers to a phenomenon that occurs when discrimination between two similar stimuli is facilitated by prior experience with those stimuli. Findings from studies with human and non- humans subjects using a short interval between presentations of two similar stimuli have shown that intermixed presentation of the stimuli results in better discrimination than presentation in a series of blocks – the Intermixed/Blocked effect (I/B effect) and suggest that stimulus comparison is critical for explaining this phenomenon. Comparison in humans has been explained in terms of better processing of the unique elements due to a process of short-term habituation (Dwyer et al., 2011). This processing bias would facilitate the formation of a better memory representation of the stimulus. It is expected that a better represented stimulus (more unified) would not only become more discriminable but will also suffer a loss of effective salience. In order to test this last prediction we ran an experiment with rats as subjects with a taste aversion paradigm, using a short interval between exposures to the stimuli to facilitate the comparison process. The results showed that there was a reduction in effective salience of the unique elements after intermixed preexposure in comparison with blocked preexposure, unlike what is usually found in animals when using the standard procedure involving a long interval between exposures to the stimuli. The results are discussed in terms of the different mechanisms underlying perceptual learning, which seem to depend on the details of the task.

Dwyer, D. M., Mundy, M. E., & Honey, R. C. (2011). The role of stimulus comparison in human perceptual learning: Effects of distractor placement. Journal of Experimental Psychology: Animal Behavior Processes, 37(3), 300.

Research funded by PSI2015-63737-P (MINECO/FEDER, EU, Spain).

Functional dissociation of basal ganglia circuits during the temporal rescaling of action

L. Ferguson, B. W. Balleine, & J. Bertran-Gonzalez Decision Neuroscience Laboratory, UNSW

Animals and humans show a remarkable capacity to adjust the timing of their actions to match the temporal requirements of the environment. Prior evidence indicates that cortico-basal ganglia circuitry is involved in the timing of actions, and may coordinate the temporal rescaling of previously

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD consolidated actions throughout learning. Here we designed a novel instrumental conditioning paradigm by which mice made self-paced adjustments to the timing of their action sequences in order to meet progressively longer instrumental requirements. We found that during early phases of training, the time between each individual action in a sequence was compressed, but as instrumental requirements progressed, animals prolonged the average duration of each sequence. We used a high- throughput reconstruction of signalling activity in large ensembles of neurons to identify regions in the basal ganglia most sensitive to temporal adjustments in this task. Remarkably, we found that animals adjusting their action timing showed a reduction of signalling activity in the posterior dorsal striatum (pDStr), which correlated with increased molecular activity in the subthalamic nucleus (STN); relative to animals not scaling the timing of their actions. We then explored the anatomical relationships between these regions using transynaptic labelling and circuit-specific tracing, and identified a shared cortico-basal ganglia projection specifically linking the motor cortex with both the pDStr and the STN. Overall, our results suggest that downstream basal ganglia circuits undergo important dissociations of molecular activity during the temporal adjustment of action, and that a shared cortico-basal ganglia pathway could be key to promoting such adaptations during learning.

Exposure to non-caloric sweet taste - exploring possible underlying mechanisms

Marta Gil1, Geoffrey Hall2, Isabel de Brugada3 1. Universidad Internacional de La Rioja (UNIR); 2. University of York, University of South Wales; 3. Universidad de Granada

Various studies conducted in our laboratory have shown that exposure to a sweet taste both with or without post-oral consequences can alter subsequent learning when such a substance is used as the unconditioned stimulus (US). The usual — but not only — explanation for this US preepxposure effect involves blocking mechanisms based on context-calories or taste-calories associations. However, these studies have also generated different outcomes depending on the motivational state of the animals, and we have suggested that (due to the effects observed when using substances without caloric properties) rather different mechanisms might be playing a role when observing the effects of preexposure to non-nutritive sweet tastes. In the review presented here, we explore two main alternatives for explaining the US preexposure effect under these circumstances. In particular, we will examine the evidence suggesting that extensive exposure to a non-nutritive sweet taste could lead to: 1) habituation of the sensory (sweet taste) properties of the substance, and 2) extinction of a pre- existing (innate) sweet-calorie association. Finally, we will discuss the potential implications of this line of research for explaining the detrimental effects of overconsumption of artificial sweeteners in our current “obesogenic environment”.

Research funded by PSI2015-63737-P (MINECO/FEDER).

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Individual differences in how learning is expressed

Adela F. Iliescu, Dominic M. Dwyer, & Robert C. Honey Cardiff University

Individual differences in conditioned responding emerge when a conditioned stimulus (CS) is associated with an unconditioned stimulus (US). Some rats predominantly interact with the CS (sign- trackers), while others predominantly approach the location in which the US (e.g., food) is about to be delivered (goal-trackers). It has been suggested that sign-trackers are more susceptible to the value of reinforcement cues and their pattern of behavior shares characteristics with addictive behavior (e.g., being impulsive and inflexible). In a series of experiments, we study the behavior of these two phenotypes (sign- and goal-tracking rats) by manipulating the relationship between the CS and US (e.g., reversing reinforcement contingencies, changing the value of the US). Our results show that behavior oriented to US was more sensitive to contingency changes than was CS-oriented behavior and, critically, this was the case in both sign-trackers and goal-trackers. The applicability of general process models of learning to individual differences in Pavlovian conditioning is evaluated.

The Role of Extinction Learning Processes in Callous-Unemotional Traits

Lindsay J Kemp and Caroline Moul University of Sydney

Emotional or behavioural responses frequently return (relapse) after an individual learns to inhibit these responses (i.e. experiences extinction learning). While there is considerable research into the prevention of relapse in the context of anxiety and addiction, it is unknown whether deficits in the relapse of behaviour may cause any psychopathology or particular trait profile. One profile that may be linked to such a process are Callous-Unemotional (CU) Traits. These traits describe individuals with callous disregard for others’ emotions, and individuals with CU traits show inflexibility in learned reward seeking behaviour. It is not known whether this inflexibility extends to learned inhibition of reward seeking behaviour, making it less prone to relapse effects. We therefore hypothesise that CU traits will be correlated with reductions in relapse of responding after extinction learning, indicating inflexible inhibitory learning. We have developed a measure of extinction of instrumental reward seeking behaviour, and later relapse of extinguished responding. Results indicate that CU traits in an undergraduate sample (N = 48) are associated with significant reductions in extinction learning. Furthermore, analysis of relapse behaviour is underway and will soon be complete. This study constitutes a novel examination of the role of relapse processes in personality.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

A Developmental Effect on Latent Inhibition: A replication and extension of Kaniel and Lubow (1986)

Ian McLaren, Rosamund McLaren and Ciro Civile University of Exeter

In 1986 Kaniel and Lubow reported an analogue of the conventional latent inhibition effect in animals that was specific to young children (aged 5). This is one of the few such demonstrations to use more or less simple exposure to a neutral stimulus and then demonstrate slower learning to that stimulus that cannot easily be interpreted as habituation or some other artefact. Replications of this effect, however, have been hard to come by. Here we report one such attempted replication that does provide support for this earlier demonstration, in that we have evidence of slower learning in those under 5 years old, and little effect above that age. We follow this up with an extension that looks at whether the effect can truly be said to be latent inhibition, or whether it is better interpreted as an attentional effect analogous to negative priming.

The Effect of Additional Negative Evidence on Generalisation

David W. Ng, Jessica C. Lee, Brett K. Hayes, Peter F. Lovibond University of New South Wales

Generalisation is the process of using past experiences to respond to novel situations. Generalisation usually involves integrating sources of positive evidence (CS+; stimulus followed by an outcome) as well as negative evidence (CS-; stimulus followed by a non-outcome). Associative models of generalization predict that the addition of negative evidence can only reduce generalized responding. Interestingly, evidence from studies of inductive reasoning suggest that negative evidence can increase generalisation. This study tested whether multiple sources of negative evidence can increase generalisation around the CS+. Three predictive learning experiments were conducted in which participants were conditioned to an aqua-coloured shape as a CS+ and two similarly coloured shapes as CS-s (double negative group). A single cue control group was only conditioned to the CS+. Across three experiments, only one experiment showed evidence of additional negative evidence increasing generalisation around the CS+. Although this higher peak was at odds with Blough’s (1975) shared element theory of generalisation, an extended version of Blough’s model that applied cue competition to multiple stimulus dimensions (i.e. shape and colour) successfully accounted for the data. These results have theoretical implications for associative theories of stimulus generalisation and the potential role of cognitive processes.

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD

Pitfalls in the Use of Regression Analysis to Investigate Unconscious Conditioning

David R. Shanks1, Simone Malejka1, and Miguel A. Vadillo2 1. University College London; 2. Universidad Autónoma de Madrid

Many researchers claim that people can detect regularities in their environment and adapt behaviour accordingly in the absence of awareness. The presumed unconscious effect of subtle cues on behaviour has been shown in a variety of settings (e.g., repetition priming, subliminal perception, contextual cuing). In a recent study, Greenwald and De Houwer (2017) apparently demonstrated the existence of unconscious conditioning—defined as greater accuracy in rapidly classifying an unfamiliar word (US) into its semantic category after seeing a predictive cue (a masked string of consonants; CS). Evidence that conditioning was indeed unconscious requires accepting the null hypothesis that participants were unaware of any contingency because the conditioned stimuli were visually imperceptible. Unfortunately, null-hypothesis significance testing is a poor method for proving the absence of an effect. Alternatively, one can regress the conditioning measure onto the visibility (awareness) measure, so that a non-zero intercept would indicate successful conditioning without awareness. However, the relationship between predictor and criterion is biased by their respective reliabilities. In particular, ignoring measurement error in the predictor will bias the regression slope towards zero, which in turn could raise a true intercept of zero above zero. Therefore, Greenwald and De Houwer employed a correction method proposed by Klauer, Draine, and Greenwald (1998), which hinges on several assumptions about the predictor’s distribution and (error) variance. In a simulation study, we show that a true intercept of zero can be overestimated if these assumptions are violated. This leads us to question the existence of an unconscious conditioning effect.

Profiling Effects of Chronic Caloric Restriction on Defensive Stress Responsivity

Matthew Zelko1,2, Stephen Robinson1, Elisa Hill-Yardin1, Helen Nasser2,3, Antonio Paolini3 1.School of Health & Biomedical Sciences, RMIT University; 2. Florey Institute of Neuroscience and Mental Health; 3. Institute of Social Neuroscience

Abstract: Chronic, mild calorie restriction (CR) reduces defensive stress responsivity in rodents (Kenny, Dinan, Cai, & Spencer, 2014; Levay, Govic, Penman, Paolini, & Kent, 2007). This reduction has been attributed to greater cardiac adaptability (Kishi, Hirooka, Ogawa, Konno, & Sunagawa, 2011; Stein et al., 2012) which may occur via reduced sympathetic activity and a shift toward parasympathetic dominance, although the biological mechanism for this effect remains unclear. To determine if the reductions in defensive stress responses is associated with parasympathetic dominance, the present study correlated behavioural responses with heart rate variability during anxiety tests and classical fear conditioning in rodents. In order to measure heart rate variability, 14 adult male Long Evans rats were implanted with cardiac telemetry devices. Subsequently, 7 of these rats had their amount of accessible food restricted to 75% of the intake of weight-matched counterparts for 3 weeks. All rats then underwent testing for anxiety-like behaviours and fear conditioning whilst heart rate was being remotely monitored. During testing, CR animals displayed lower levels of anxiety-like behaviours, as measured in the elevated plus maze and open field test, as well as a shorter freezing duration during fear extinction. These behavioural changes were correlated with greater power in the high frequency

30th Anniversary International Australian Learning Group Conference 24 – 26th July 2019 Magnetic Island QLD component of heart rate, which is indicative of parasympathetic dominance. These findings confirm that CR produces greater behavioural adaptability to stress which is accompanied by a shift toward parasympathetic dominance of the autonomic nervous system.

Event and temporal uncertainty in Pavlovian conditioning

Jorge Mallea1,2 & Peter Balsam1,2,3 1.Department of Psychology, Columbia University; 2. New York State Psychiatric Institute; 3.Department of Psychology, Barnard College

Pavlovian stimuli decrease uncertainty about what is going to happen in the future. The co- occurrence of events in the world is not usually perfect and there is uncertainty in most situations. Two types of uncertainty are event and temporal uncertainty. Event uncertainty in Pavlovian conditioning is uncertainty about whether the unconditioned stimulus (US) will come in that particular trial. Temporal uncertainty refers to uncertainty about when the US will appear if, in fact, it does occur on that particular trial. The aim of the present experiment is to study the behavioral effects of event and temporal uncertainty in learning and extinction of Pavlovian associations. For this, four groups of mice were exposed to appetitive Pavlovian conditioning in which a light conditioned stimulus (CS) predicted the occurrence of a reinforcer. In different groups, the probability of the US and the variability of the CS duration were manipulated. During acquisition, two groups were reinforced after 100% of the trials (Fixed-100 and Variable-100) and two of them after 33% of the trials (Fixed-33 and Variable-33) and trained with Fixed or Variable CS durations. Results show that both event and temporal uncertainty reduce the level of conditioned responding, increasing latencies to respond, and decreasing time spent on the magazine and proportion of trials with responses. The implications of this for theories of conditioning and its implication for understanding neural mechanisms for processing uncertainty during Pavlovian learning are discussed