Research Article: New Research | Cognition and Behavior Perirhinal and postrhinal damage have different consequences on attention as assessed in the five-choice serial reaction time task https://doi.org/10.1523/ENEURO.0210-21.2021
Cite as: eNeuro 2021; 10.1523/ENEURO.0210-21.2021 Received: 11 May 2021 Accepted: 27 August 2021
This Early Release article has been peer-reviewed and accepted, but has not been through the composition and copyediting processes. The final version may differ slightly in style or formatting and will contain links to any extended data.
Alerts: Sign up at www.eneuro.org/alerts to receive customized email alerts when the fully formatted version of this article is published.
Copyright © 2021 Trettel et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license, which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed.
1 Perirhinal and postrhinal damage have different consequences 2 on attention as assessed in the five-choice serial reaction time task 3 4 Sean G. Trettel, Kara L. Agster and Rebecca D. Burwell 5 Brown University Departments of Psychology and Neuroscience 6 7 Abbreviated Title: Perirhinal and postrhinal cortices and attention 8 9 Corresponding author: 10 Rebecca D. Burwell 11 Department of Cognitive, Linguistic, and Psychological Sciences 12 Brown University 13 190 Thayer Street 14 Providence, RI 02912 15 [email protected] 16 401-863-9208 17 18 Other authors: 19 Sean G. Trettel 20 Department of Cognitive, Linguistic, and Psychological Sciences 21 Brown University 22 190 Thayer Street 23 Providence, RI 02912 24 [email protected] 25 401-863-9209 26 27 Kara L. Agster 28 University of Colorado Boulder 29 552 UCB 30 Boulder, CO 80309-0552. 31 [email protected] 32 303-735-2927 33 34 Number of figures and tables: 2, 7 35 Number of Pages: 34 36 Number of words for Abstract: 249 37 Introduction: 481 38 Discussion: 1115 39 40 Key words (6): parahippocampal, hippocampus, medial temporal lobe, memory, serial 41 reaction time, executive function 42 43 The authors declare no competing financial interests. 44 45 Acknowledgements: This research was supported NEI F32-EY031561 to SGT, NIMH 46 F31-MH072144 to KLA, and NIMH R01-MH108729, NSF IBN9875792, and NSF IOS- 47 1656488 to RDB. 48 49 50
1
51 Abstract 52 The perirhinal (PER) and postrhinal (POR) cortices, structures in the medial temporal 53 lobe, are implicated in learning and memory. The PER is understood to process object 54 information and the POR to process spatial or contextual information. Whether the 55 medial temporal lobe is dedicated to memory, however, is under debate. In this study, 56 we addressed the hypothesis that the PER and POR are also involved in non-mnemonic 57 cognitive functions. Rats with PER or POR damage and SHAM surgical controls were 58 shaped, trained, and tested on the five-choice serial reaction time (5CSRT) task, which 59 assesses attention and executive function. Rats with PER damage were impaired in 60 acquiring the task and at asymptote, even though processing information about objects 61 was not relevant to the task. When confronted with attentional challenges, rats with PER 62 damage showed a pattern consistent with decreased attentional capacity, increased 63 response errors, and increased impulsive behavior. Rats with POR damage showed 64 intact acquisition and normal asymptotic performance. They also exhibited faster 65 latencies in the absence of speed accuracy trade-off suggesting enhanced response 66 readiness. We suggest this increased response readiness results from decreased 67 automatic monitoring of the local environment, which might normally compete with 68 response readiness. Our findings are consistent with a role for PER in controlled 69 attention and a role for POR in stimulus-driven attention providing evidence that the PER 70 and POR cortices have functions that go beyond memory for objects and memory for 71 scenes and contexts. These findings provide new evidence for functional specialization 72 in the medial temporal lobe. 73 74 Significance 75 The perirhinal (PER) and postrhinal (POR) cortices, structures in the medial temporal 76 lobe, are implicated in learning and memory. Whether medial temporal lobe structures 77 are exclusively dedicated to memory, however, is under debate. We provide evidence 78 for a role for PER in controlled attention and a role for POR in stimulus-driven attention. 79 These findings provide new evidence for functional specialization in the medial temporal 80 lobe. 81
82 Key words (6): parahippocampal, hippocampus, medial temporal lobe, memory, serial 83 reaction time, executive function 84
2
85 Introduction
86 Episodic memory, or memory for the everyday events of life, is understood to be
87 supported by the medial temporal lobe (Squire et al., 2004; Eichenbaum et al., 2007). In
88 primates, the medial temporal lobe comprises the hippocampal formation and the nearby
89 structures of the parahippocampal region including the perirhinal cortex (PER) and the
90 parahippocampal cortices, the postrhinal cortex (POR) in the rodent brain (Burwell et al.,
91 1995; Beaudin et al., 2013). The PER is thought to process object information and the
92 POR to process spatial or contextual information in the service of episodic memory.
93 Whether all medial temporal lobe structures are dedicated exclusively to
94 mnemonic functions remains under debate. A number of studies have provided evidence
95 that different structures in the medial temporal lobe make different contributions to
96 memory and learning (Bussey et al., 1999; Jarrard et al., 2004; Eacott and Gaffan, 2005;
97 Eichenbaum et al., 2007; Ramos, 2013; Heimer-McGinn et al., 2017). There is evidence
98 for functional specialization along the septotemporal axis (Maurer and Nadel, 2021).
99 Some investigators have proposed that components of the medial temporal lobe
100 memory system are differentiated by what is represented and how information is
101 processed (Jarrard et al., 2004; Bussey and Saksida, 2007; Lawrence et al., 2020).
102 Finally, there is evidence that some medial temporal lobe structures have non-mnemonic
103 functions including perception (Bucci and Burwell, 2004; Bussey and Saksida, 2007; Ahn
104 and Lee, 2017; Gaynor et al., 2018). We previously provided evidence that the POR has
105 a role in attention (Burwell and Hafeman, 2003; Bucci and Burwell, 2004). In this study,
106 we addressed the hypothesis that PER and POR support different aspects of attention in
107 the service of learning and memory as well as other cognitive processes.
108 We have proposed that the POR represents context by combining spatial
109 information with information about discrete items and objects provided by the PER
110 (Furtak et al., 2012; Heimer-McGinn et al., 2017). We further proposed that the POR is
3
111 involved in ongoing and automatic monitoring of the context for changes (Burwell and
112 Hafeman, 2003). In contrast, the PER is involved in processing individual objects and
113 items, especially when such stimuli are complex or ambiguous (Cowell et al., 2010).
114 Such findings suggest the POR and PER may be involved in different sorts of attention,
115 i.e. stimulus driven and controlled attention, respectively.
116 In the present study, we used the 5-choice serial reaction time (5CSRT) task to
117 address the hypothesis that the PER and POR functions include different types of
118 attention. The 5CSRT task is a powerful tool for assessing multiple dimensions of
119 attention, including controlled, sustained, divided, selective, and visuospatial attention
120 (Carli et al., 1983; Winstanley et al., 2003; Bari et al., 2008). Rats with damage to either
121 PER or POR were shaped and trained to asymptote on the task. They were then
122 presented with a series of attentional challenge conditions including noise distractors,
123 shortened stimulus durations, variable intertrial intervals, and a random tone distractor.
124
125 Materials and Methods
126 Subjects
127 Subjects were 28 male Long-Evans rats (Charles River Laboratories, Wilmington,
128 MA) weighing between 250-300 grams at the start of the experiments. Animals were
129 allowed a few days to adjust to the animal colony. At that point, we began handling the
130 animals daily. After approximately one week, the animals were put on a food schedule
131 such that one meal a day was provided with the goal of maintaining body weight at 85-
132 90% of the free feeding weight at the same age. Water was freely available throughout
133 the experiment. Animals were maintained on a 12:12 hour reverse light:dark cycle. Thus,
134 all testing occurred during the dark period of the light cycle. This research was carried
135 out according to Brown institutional and federal animal care and use guidelines.
136 Surgery
4
137 Animals were placed in an anesthesia induction chamber and lightly anesthetized
138 with isoflurane gas. The scalp was shaved and subjects were placed in a stereotaxic
139 apparatus (David Kopf Instruments, Tujunga, CA). All surgical procedures were carried
140 out under isoflurane anesthesia. Animals were also given glycopyrrolate (0.25 mg/kg,
141 IP) to reduce the occurrence of respiratory difficulties. Once fully anesthetized an
142 incision was made over the top of the skull; the skin and fascia were retracted. The skull
143 overlying the postrhinal or perirhinal cortex was removed. A microinjection unit (David
144 Kopf Instruments, Tujunga, CA) was used to deliver either a quantity of ibotenic acid
145 (lesion cases) or sterile saline (sham controls). A volume of 50 nL of ibotenic acid (10
146 µg/1µL) was injected to each site in the perirhinal cortex and a volume of 75 nL was
147 injected to each site in the postrhinal cortex. The micropipette was lowered to the
148 appropriate coordinate and allowed to sit for 1 minute; following delivery of the
149 excitotoxin or saline, the micropipette was left in place for 2 minutes and then slowly
150 raised and removed from the brain. After all injections were made the skull was cleaned
151 and the wound sutured. Animals were then returned to their home cage and placed
152 under a warming light. Once active, subjects were given a dose of rimadyl (Bio-serv, 6
153 mg, PO) to relieve pain. Animals were then returned to the colony, and observed for 3
154 days to ensure proper healing of the wound. Subjects were allowed to recover for a
155 period of 2 weeks before starting behavioral testing. Table 1 shows the lesion
156 coordinates for PER lesioned subjects (n=8) and POR lesioned subjects (n=8). One half
157 of sham controls (n=6) received saline injections at PER locations, and the other half
158 (n=6) received saline injections at POR locations.
159 Histology
160 At the completion of the experiment animals were given a lethal overdose of
161 Beuthanasia (Schering-Plough, Kenilworth, NJ) and transcardially perfused. First,
162 normal saline was perfused to clear the blood. Subsequently, 10% formalin was
5
163 perfused to fix the tissue. Subjects were decapitated and brains were removed for
164 histological processing. Brains were cryoprotected in a 20% glycerin solution and
165 subsequently sectioned on a freezing microtome. We collected a 1:2 series of coronal
166 sections with a thickness of 40 μm. Sections were mounted on gelatin subbed slides
167 and allowed to dry overnight in an oven (40°C). Slides were defatted for 60 min in a
168 50% ethanol/50% chloroform solution, rehydrated through a descending series of
169 alcohols to rehydrate the tissue (100%, 95%, 75%, and 50% ethanol), stained with
170 thionin solution, dehydrated through an ascending series of alcohols to dehydrate the
171 sections, placed in a xylenes series, and coverslipped.
172 Lesions were quantified using the Cavalieri method as in earlier reports (Heimer-
173 McGinn et al., 2017). Briefly, every other 40 μm section was examined at 4x
174 magnification and regions of interest were drawn. Regional borders were marked off
175 and damaged areas were highlighted. Damaged tissue both inside and outside the
176 region of interest was quantified. Tissue that exhibited cell death, damage, or cortical
177 thinning was considered functionally damaged. For each coronal section, we quantified
178 the total area of the target region, the area of the target region that was damaged, and
179 the area that was damaged in non-target regions for the left and right hemispheres. We
180 then calculated the percent of the target area that was damaged, and the percent of the
181 coronal sections that exhibited any amount of damage. In earlier studies, we have found
182 that the rostrocaudal extent of the damage is a better predictor of lesion effect than area
183 or volume. For that reason, subjects were retained in the study if the lesion was
184 distributed along the rostrocaudal axis in both hemispheres. In addition, subjects with
185 bilateral damage to an extra-target region were eliminated from the study.
186 Apparatus
187 Behavioral training was conducted in an operant testing environment controlled
188 by the MED-PC software package (MedAssociates, Inc., Burlington, VT). Custom
6
189 software written in the Pascal-based, MED-PC notation controlled the behavioral tasks
190 and recorded task events and responses. Experiments were conducted in six
191 24.0×30.5×29.0 cm operant test chambers with aluminum panels in the front and back,
192 Plexiglas side walls and top, and a grid floor. A dimly illuminated food cup was recessed
193 in the center of one end wall. A photo beam mounted inside the recessed food cup
194 permitted automated assessment of food-cup behavior. The opposite wall was curved
195 and incorporated 5 nose poke holes with LED stimulus lights inside. A partially-shaded
196 house light was mounted centrally at the top of the food cup wall and provided
197 background lighting throughout the session. Each testing chamber was enclosed in a
198 62.0×56.0×56.0 cm sound-attenuating chamber fitted with an exhaust fan that provided
199 air flow to the test chamber and background white noise.
200 Behavioral Procedures
201 Task: The start of a session was signaled by illumination of the house light and
202 delivery of one food pellet (45 mg, Bio-Serv, Frenchtown, NJ). Retrieval of the pellet
203 activated the head entry detectors at the reward port and served to initiate the first trial.
204 Following a fixed intertrial interval (ITI) of 5 seconds, one stimulus port was illuminated.
205 Animals were required to nose poke in the illuminated port to indicate a correct
206 response. While the stimulus port was lit and for a short period afterwards, termed the
207 limited hold period, nose poke responses in the port were rewarded with delivery of a
208 pellet. Nose poke responses in ports that were not illuminated were defined as incorrect
209 responses. Errors of omission were defined as a failure to respond during the limited
210 hold period. Both incorrect responses and omission errors were punished with a time
211 out - house light off for 5 seconds. Nose poke responses in any of the stimulus ports or
212 the reward port restarted the time out period. Subsequent trials were initiated by head
213 entry to the reward port or completion of the time out period. Perseverative responses
214 were defined as nose pokes following the initial correct response and were punished by
7
215 a time out. Premature responses were defined as nose pokes made during the ITI and
216 required the animal to initiate a trial by responding in the reward port. Daily testing
217 sessions consisted of 100 trials or 30 minutes of testing, whichever occurred first. At the
218 completion of the session the house light was extinguished and responses were no
219 longer rewarded.
220 Shaping and Training: Initially stimulus ports were illuminated for long periods to
221 facilitate learning. At the start of shaping stimulus ports were illuminated for 64 seconds.
222 Once subjects reached a criterion performance of 75% correct for two consecutive
223 sessions, the stimulus duration was decreased to the next level. Shaping stimulus
224 durations (limited hold) were 64, 32, 16, 8, 4, 2, 1.5, and 1 seconds. If subjects were not
225 able to reach a criterion of 75% within 10 sessions, they were advanced to the next
226 phase. Once shaping was complete, subjects completed 20 days of training in the
227 standard condition with the 0.5 s stimulus duration.
228 Attentional Challenges: Following completion of 20 days training on the standard
229 condition, a series of behavioral challenges were introduced. Subjects completed four
230 challenge sessions, interspersed with baseline sessions (standard condition). Challenge
231 conditions included noise distractor presented with the target, shortened stimulus
232 durations, variable ITIs, and tone distractor presented during the ITI. In the noise
233 challenge, a burst of white noise was introduced immediately prior to illumination of the
234 stimulus port. In the Short challenge, stimulus durations were shortened from 0.5
235 seconds to 0.25 seconds. The third challenge condition introduced variable ITIs instead
236 of the constant 5 second ITI. Variable ITIs included 1, 3, or 5 seconds. Finally, in the
237 tone challenge, a tone was introduced at random periods during the standard ITI.
238 Data Analysis
239 Shaping, training, and attentional challenge phases were analyzed separately by
240 univariate and repeated-measures analysis of variance (rANOVA). We were interested
8
241 both in whether the POR and PER groups differed from the SHAM group and in whether
242 POR and PER groups differed from each other, regardless of whether there was a main
243 effect of lesion group. Thus, for between group analyses, we conducted planned
244 comparisons of the POR and PER lesion groups to the SHAM group and to each other.
245 For analyses of shaping, stimulus duration Level was the within-subject variable.
246 The dependent variables analyzed included sessions to criterion (STC), accuracy
247 (number correct/(number correct + number incorrect) * 100), omissions, premature
248 responses, perseverative responses, and head entries at the food port. In addition, we
249 analyzed latencies for correct and incorrect responses as well as latency to respond at
250 the food port (correct latency, incorrect latency, and reward latency).
251 For analysis of training, the twenty sessions were grouped into blocks of five
252 sessions, and block was the within-subject variable. The dependent variables included
253 accuracy, omissions, premature responses, perseverative responses, head entries,
254 correct latency, incorrect latency, and reward latency.
255 For attentional challenges, challenge sessions were alternated with standard
256 baseline sessions and performance on a challenge was compared with the prior
257 baseline session. We first analyzed each group individually on each challenge type with
258 Condition as the within-subject variable. We then conducted planned comparisons of the
259 POR and PER lesion groups to the SHAM group and to each other, similar to the
260 analyses of shaping and training except that condition was the within-subject variable
261 rather than session or block. Finally, in order to better assess the responses to
262 attentional challenges, we calculated a Challenge Ratio, CR = (Challenge-
263 Baseline)/(Challenge+Baseline), such that scores near zero indicate little change from
264 baseline during the challenge. Positive and negative scores indicate increases and
265 decreases from baseline, respectively. The Challenge Ratio allows the reader to quickly
266 see whether the challenge performance was higher or lower than baseline.
9
267 All analyses were conducted using SAS version 8.0 or 9.1 (SAS Institute, Cary,
268 NC) or SPSS version 24 (IBM, Armonk, NY). A significance level of 0.05 was used for all
269 analyses.
270 Results
271 Histology
272 Three groups of animals were tested on the 5CSRT task: POR (n=8), PER (n=8),
273 and sham surgery controls (SHAM, n=12). Examination of coronal sections through the
274 POR for each lesioned subject revealed that tissue was damaged throughout the
275 rostrocaudal extent of the region. A total of 94.5% of coronal sections exhibited some
276 amount of damage. Using the Cavalieri method the average volume of POR damage
277 was estimated at 36.9%. There was a very small amount of damage to the bordering
278 medial entorhinal cortex in all POR cases, but the damage was small and predominantly
279 unilateral. PER lesion analysis also revealed tissue damage along the entire
280 rostrocaudal extent of the region. A total of 88.8% of coronal sections exhibited some
281 amount of obvious damage. The average volume of PER damage was estimated at
282 44.9%. In seven PER cases there was some damage limited to the most lateral part of
283 the ventrally adjacent entorhinal cortex, and in five cases there was a small amount of
284 damage to the dorsally adjacent temporal association cortex. Schematics of the largest
285 and smallest lesions to the POR and PER are presented in Figure 1.
286 Figure 2 shows examples of the type of damage that that was assessed in our
287 histological analysis. Damage was identified as missing cortex, thinning cortex, missing
288 cells, and pyknotic cells. Pyknotic cells appear as smaller and darker than healthy cells.
289 One section from a PER subject (Figure 2A) shows apparent damage to the superficial
290 layers and middle layers, including superficial layer V in the dorsoventral extent of the
291 PER. Layer I is thinner than in control subjects. There are areas in which cells have
292 disappeared, and there are other areas in which cells have become pyknotic. There is
10
293 likely secondary damage to cells in deep layers that project to the damaged superficial
294 layers. Additionally, the PER has intrinsic longitudinal connections which would also
295 contribution to secondary damage not likely to be apparent in cell stains. Figure 2B
296 shows a section from the brain of a POR subject. In this case, damage to deep layers
297 presents mainly as pyknotic cells. Additionally, there is thinning of all cortical layers.
298 Again, secondary damage would be expected in superficial layers due to dying cells in
299 deep layers, and we would also expect that there is due to intrinsic connections. In this
300 case, there is some damage in dorsally adjacent ventral temporal cortex. There is
301 possibly some layer II damage in the ventrally adjacent lateral entorhinal area, but most
302 of the open space is actually the lamina dissecans and likely not missing cells.
303 Two questions of interest are 1) whether lesion size correlates with performance,
304 and 2) whether damage outside of the target regions (PER and POR) can account for
305 our results. To address these questions, we further analyzed our histological data. As
306 described in the methods, lesion size was quantified by the percentage of the region that
307 was damaged, and lesion distribution was quantified by the percentage of coronal
308 sections that showed any damage. Lesion size and lesion distribution were quantified
309 separately for target region damage and for non-target region damage, yielding four
310 variables for each subject. For POR and PER groups combined, none of the four
311 variables was correlated with percent accuracy: p > 0.952 for target lesion size, p >
312 0.475 for target lesion distribution, p > 0.249 for non-target lesion size, and p > 0.149 for
313 non-target lesion distribution. There were also no significant correlations for PER
314 histology analyzed separately (p’s > 0.823). For the POR analysis neither target lesion
315 variable was significantly correlated with accuracy (p’s > 0.549). Non-target lesion size
316 and distribution were significantly positively correlated with accuracy for both variables.
317 This correlation, however, was due to a single POR subject that exhibited by far the
318 largest non-target region damage and also exhibited the highest accuracy of the group.
11
319 Without that animal the POR correlation analyses were not significantly correlated with
320 accuracy: p > 0.845 for non-target lesion size, and p > 0.907 for non-target lesion
321 distribution. We also ran a non-parametric test, the Independent-Samples Mann-Whitney
322 U Test, for group differences between the PER and POR groups for all four lesion
323 variables. There was no significant group difference on any variable: target lesion size,
324 p>0.195; target lesion distribution, p>0.234: non-target lesion size, p>0.798; non-target
325 lesion distribution, p>0.878. Thus, neither lesion size nor extra-target region damage can
326 account for our results.
327 Shaping
328 Subjects were initially tested with a stimulus duration of 64 seconds, and
329 progressively shaped down to a stimulus duration of 1 second (Figure 3A). A
330 performance criterion for advancing to the next stimulus duration was set at > 75%
331 accuracy for 2 consecutive days. However, beginning with the 32 s duration if the
332 criterion had not been met in 10 sessions, the subject was progressed to the next
333 stimulus duration. All of the POR subjects reached criterion on each stimulus duration in
334 less than 10 daily sessions. Seven of eight PER subjects were advanced for at least one
335 stimulus duration. Three SHAM subjects were advanced for one or both of the shortest
336 stimulus durations. Overall, the POR group required fewer sessions (24.00 ± 1.28) and
337 the PER group required more sessions (55.13 ± 9.48) as compared to the SHAM group
338 (35.85 ± 3.96) to reach criterion (Figure 3A). The PER group showed significantly higher
339 sessions to reach criterion compared with the POR group (F(1,14) = 10.58, p < 0.006) and
340 with the SHAM group (F(1,17) = 7.37, p < 0.015). The POR and SHAM groups were not
341 different from each other (p > 0.16). Note that because so many PER rats and a few
342 SHAM rats were advanced without reaching criterion of 75% accuracy, their training and
343 baseline performances tend to be lower than criterion.
12
344 In addition to slow acquisition during shaping, the PER group was impaired on a
345 number of measures compared with the SHAM group (Figure 3). This was evident in
346 significant main effects of group and / or group x shaping level interactions (Figure 3A-C,
347 F). The PER group exhibited significantly lower percent accuracy (Figure 3B; F1,17=9.7,
348 p<0.007), which improved slightly across shaping (F7,119=2.936, p<0.008) and
349 significantly higher percent omissions that worsened slightly across shaping (Figure 3C;
350 F7,119=2.94, p<0.007). Finally, early in shaping, the PER group showed higher latencies
351 to respond after incorrect trials, but latencies were similar to those of the SHAM group by
352 the end of shaping (Figure 3D; F7,119=3.125, p<0.005).
353 The PER group was also significantly impaired on a number of measures as
354 compared with the POR group (Figure 3A-F). The PER group exhibited significantly
355 lower percent accuracy (Figure 3B; F1,14=8.44, p<0.012) and significantly higher percent
356 omissions (Figure 3C; F1,14=12.18 p<0.005). The PER group exhibited a trend towards
357 fewer premature responses than the POR group (Figure 3E; F1,14=4.15, p<0.062). Also
358 similar to the SHAM group, the PER group showed higher latencies to respond than the
359 POR group after incorrect trials, but latencies were similar to those of the SHAM group
360 by the end of shaping (Figure 3E; F7,98=3.35, p<0.031).
361 The POR and SHAM groups were not different from each other except on
362 perseverative responses (Figure 3F). There was no main effect of group but there was a
363 group by level interaction (F7,119=4.40, p<0.0003). Numerically, the POR group had more
364 perseverations at limited holds of 64, 4, and 2 seconds and fewer perseverations at 32,
365 16, and 8 seconds. Post hoc analyses indicated that the POR group had significantly
366 more perseverations at the 64 s level of shaping (p<0.021) and marginally significantly
367 less perseverations at the 8 s level of shaping (p<0.077).
368 Training
13
369 After reaching criterion during shaping, all subjects were trained on the standard
370 task (stimulus duration of 0.5 seconds) for 20 days. One SHAM subject that was
371 performing at 73.6% accuracy at the end of shaping dropped to a mean accuracy across
372 training of 23.2%. This subject was removed from all analyses. Due to experimenter
373 error, data files for training day 13 were lost for all but two subjects. These data were
374 estimated for each animal by taking the mean of days 12 and 14.
375 Four blocks of five sessions were analyzed for the variables of interest, including
376 accuracy, omissions, premature responses, perseverative responses, latency to
377 respond, and latency to retrieve reward. Planned comparisons indicated no group by
378 block interactions in performance in the standard task. There were, however, a few
379 group differences for accuracy (Figure 4A, B) and latencies to respond on correct trials
380 (Figure 4C, D). On accuracy, the two lesion groups were not different from the SHAM
381 group but were different from each other in that the PER rats were impaired compared to
382 the POR rats. The PER group showed marginally significantly lower accuracy compared
383 with the SHAM group (Figure 4B; F(21,17) = 3.94, p < 0.06). The POR group showed
384 similar accuracy compared with the SHAM group (p > 0.39) and significantly higher
385 accuracy compared with the PER group (F(1,14) = 8.79, p < 0.01). The POR group
386 showed marginally lower latencies on correct trials compared with the SHAM group
387 (Figure 4D; p < 0.052) and significantly lower latencies compared with the PER group
388 (F(1,14) = 7.24, p < 0.018). Other than accuracy and latencies on correct trials, planned
389 paired comparisons revealed no other significant between-group differences.
390
391 Responses to attentional challenges
392 Within-subject effects. We first analyzed responses to attentional challenges for
393 each group, individually, to determine whether there was a significant impact of each
394 challenge on each performance measure except head entries. The measure of head
14
395 entries showed very high variance. Upon inspection of the data, it was determined that
396 the number of head entries in some session for some subjects was impossibly high.
397 Since we were unable to determine a reasonable correction, head entries were not
398 subjected to further analysis. One SHAM subject was inadvertently not run on the last
399 challenge in which a tone was presented during the ITI.
400 Starting with the SHAM group, the Variable ITI condition in which the ITI was 1,
401 3, or 5 seconds instead of the constant 5 seconds, had the greatest impact on
402 performance (Table 2, top). Accuracy was significantly lower (F(1,10) = 10.35, p < 0.009),
403 and omissions were significantly higher (F(1,10) = 11.42, p < 0.007). Latencies were also
404 significantly different with slower latencies for both Correct (F(1,10) = 5.07, p < 0.05) and
405 Incorrect Trials (F(1,10) = 37.26, p < 0.001), and faster latencies to approach the Reward
406 port (F(1,10) = 8.16, p < 0.02). For the Short condition (target shortened to 250 msec),
407 accuracy was also significantly lower (F(1,10) = 23.95, p < 0.001). Introducing a burst of
408 white noise immediately prior to presentation of the visual target (Noise condition)
409 significantly increased omissions (F(1,10) = 7.32, p < 0.02). There was no impact of Tone
410 condition (tone during the ITI) on any measure, and there was no impact of any
411 challenge on premature responses, or perseverations.
412 The PER group appeared to be less sensitive to attentional challenges than the
413 SHAM group (Table 2, middle). Premature responses were significantly increased by the
414 noise distractor during the target (F(1,7) = 17.57, p < 0.004), significantly decreased by
415 the variable ITI (F(1,7) = 6.09, p < 0.04), and marginally significantly increased by the tone
416 presented during the ITI (F(1,7) = 4.99, p < 0.06). During the variable ITI, latencies to
417 select were significantly higher on Incorrect trials (F(1,7) = 21.80, p < 0.002), and were
418 marginally higher on Correct trials (F(1,7) = 4.68, p < 0.07). For the noise during target
419 presentation challenge, latencies to select were marginally significantly higher on
420 Incorrect trials (F(1,7) = 5.48, p < 0.0.052), and Correct trials (F(1,7) = 4.47, p <
15
421 0.072).There was no impact of the short target presentation on any measure, and there
422 was no impact of any challenge on accuracy, omissions, perseverations, or latency to
423 approach the reward port. It is surprising that shortening the stimulus duration did not
424 increase premature responding, as that is a common finding in the 5CSRT task.
425 Shortening the target duration did result in a numerical increase in premature responding
426 for all three groups when compared to the immediately prior baseline. (Note that for ease
427 of presentation in Figures 5 and 6, we showed the average across all baseline sessions
428 and not the immediately prior baseline.) The variability was such that none of these
429 increases were significant.
430 The POR group also appeared to be less sensitive to attentional challenges than
431 the SHAM group (Table 2, bottom). Only omissions and latencies were affected by
432 challenges. Omissions were significantly increased over baseline by the noise distractor
433 during the target (F(1,7) = 6.5, p < 0.04), and marginally significantly increased by the
434 shortened target (F(1,7) = 4.89, p < 0.063). For the noise during target presentation
435 challenge, latencies to select were significantly higher on incorrect trials (F(1,7) = 9.67, p <
436 0.0.02) and correct trials (F(1,7) = 7.52, p < 0.03), and the latencies to approach the food
437 port were marginally significantly higher on (F(1,7) = 4.45, p < 0.0.073). During the
438 variable ITI challenge, latencies to select were significantly higher on Incorrect trials
439 (F(1,7) = 15.04, p < 0.0.006) and Correct trials (F(1,7) = 19.01, p < 0.0.003), but the
440 latencies to approach the food port were significantly lower (F(1,7) = 8.84, p < 0.0.02).
441 There was no impact of the tone presentation during the ITI on any measure, and there
442 was no impact of any challenge on accuracy, premature responses, or perseverations.
443 Between-subject effects. Group differences were assessed separately for the
444 average of the four baselines prior to challenge sessions and for each of the challenge
445 sessions. Overall, there were more group differences in accuracy followed by percent
446 omissions, and group differences were often in the same directions as for training.
16
447 For percent accuracy (Figure 5A), the PER group showed significant decreases
448 compared to the SHAM group for average baseline (F(1,17) = 5.57, p < 0.03), Noise (F(1,17)
449 = 15.34, p < 0.03), Short (F(1,17) = 5.04, p < 0.03), and variable ITI (F(1,17) = 8.72, p <
450 0.008) conditions, and marginally significantly decreased percent accuracy for the Tone
451 condition (F(1,16) = 3.74, p = 0.0710). The PER group also showed significantly
452 decreased percent accuracy compared to the POR group (Figure 5A) for the average
453 baseline (F(1,14) = 16.52, p < 0.03), Noise (F(1,14) = 9.03, p < 0.009), Short (F(1,14) = 15.85,
454 p < 0.001), var ITI (F(1,14) = 19.97, p < 0.0005), and Tone conditions (F(1,14) = 11.06, p <
455 0.005). The POR and SHAM groups were not significantly different on accuracy of
456 performance for baseline or for any attentional challenges.
457 For percent omissions (Figure 5B), the PER group showed marginally significant
458 increases compared to the SHAM group for average baseline (F(1,17) = 3.11, p < 0.09)
459 and significant increases for the Short condition (F(1,17) = 5.68, p < 0.03). The PER group
460 also showed marginally significantly increased omissions compared to the POR group
461 for the average baseline (F(1,14) = 4.47, p < 0.053) and significant increases for the Short
462 (F(1,14) = 5.63, p < 0.03), var ITI (F(1,14) = 4.72, p < 0.04), and Tone conditions (F(1,14) =
463 5.25, p < 0.03). The POR and SHAM groups were not significantly different on accuracy
464 of performance for baseline or any attentional challenges. Interestingly, the POR group
465 showed marginally significantly decreased omissions compared to the PER group for the
466 average baseline (F(1,14) = 4.47, p < 0.05) and significant decreases for the Short (F(1,14) =
467 5.63, p < 0.03), var ITI (F(1,14) = 4.72, p < 0.04), and Tone conditions (F(1,14) = 5.25, p <
468 0.03). The POR group also showed significantly decreased omissions compared to the
469 SHAM group for the Tone condition (F(1,16) = 4.52, p < 0.05).
470 There were no significant differences for the percent trials with premature
471 responses (Figure 5C) or for percent trials with perseverative responses (data not
472 shown). The PER group, however, showed marginally significantly increased percent
17
473 premature responses compared to the SHAM group for the Noise condition (Figure 5C)
474 (F(1,14) = 4.01, p < 0.065).
475 For latencies on correct trials the POR group was faster than the SHAM group on
476 baseline and each of the challenges, and faster that the PER group on baseline and all
477 challenges except the noise challenge (Figure 6A). The POR group was significantly
478 faster than the SHAM group on average baseline (F(1,17) = 4.71, p < 0.044), the noise
479 challenge (F(1,17) = 7.38, p < 0.02), and the tone challenge (F(1,16) = 5.86, p < 0.023), and
480 was marginally significantly faster on the Short (F(1,17) = 3.96, p < 0.063), and variable ITI
481 (F(1,16) = 4.17, p = 0.057) challenges. The POR group also showed significantly faster
482 latencies on correct trials compared to the PER group for average baseline (F(1,14) =
483 12.79, p < 0.003), Short (F(1,14) = 5.62, p < 0.03), var ITI (F(1,14) = 20.837, p < 0.0004),
484 and Tone conditions (F(1,14) = 7.55, p < 0.016). For all challenges except the noise
485 challenge the pattern of group differences was similar to the pattern for the baseline. For
486 the noise challenge the pattern was somewhat different. The PER group showed faster
487 latencies compared to baseline, whereas the other two groups did not. (Figure 6A).
488 Analysis of correct latencies for the noise challenge showed significant group by
489 condition interactions for both the PER vs. SHAM comparison (F(1,17) = 5.44, p < 0.032)
490 and the PER vs. POR comparison (F(1,14) = 7.93, p < 0.0138).
491 There were fewer group differences in latencies on incorrect trials (Figure 6B).
492 The POR group was again faster than both the SHAM and the PER groups on baseline
493 (F(1,17) = 6.61, p < 0.02 and F(1,14) = 6.77, p <0.021, respectively). The POR group was
494 also significantly faster than the SHAM group on the variable ITI challenge (F(1,17) = 6.45,
495 p < 0.022), and the tone challenge (F(1,16) = 7.09, p < 0.017). On the noise challenge, the
496 POR was slower relative to baseline and PER group was faster relative to baseline. This
497 was evident in significant group by condition interaction for the PER vs. POR
498 comparison (F(1,14) = 14.58, p < 0.002).
18
499 For the latency to retrieve a food reward following a correct trial, there was only a
500 single significant result with regard to challenges. On the noise challenge, both the POR
501 and PER groups were slower relative to baseline, but the change was greater for the
502 POR group (Figure 6C). In other words, latencies slightly increase from baseline to noise
503 condition for the SHAM and PER group, the latencies dramatically increased from
504 baseline to noise condition for the POR group. This was evident in a significant group by
505 condition interaction for the PER vs. POR comparison (F(1,14) = 6.57, p < 0.023).
506 Analysis of Challenge Ratios. In order to better assess changes from baseline
507 during attentional challenges, we constructed a Challenge Ratio such that CR =
508 (Challenge-Baseline)/(Challenge+Baseline). This ratio is mathematically equivalent for
509 the discrimination ratio used for some novel object exploration and preferential viewing
510 studies. Scores near zero indicate little change from baseline during the challenge.
511 Positive scores indicate increases relative to baseline, and negative scores indicate
512 decreases relative to baseline. We have included these analyses because they better
513 control for baseline performance and they allow another way for the reader to assess
514 responses to attentional challenges.
515 ANOVA of the CR measure for group differences indicated no group differences
516 in accuracy challenge relative to baseline, although for all groups and challenges, the
517 mean accuracy was numerically lower on the challenge session compared with the
518 immediately prior baseline session (Figure 7). P values for accuracy Challenge Ratios
519 ranged from 0.31 to 0.97. There were also no group differences in the ratios for
520 perseverative responses. All perseveration ratios were close to zero, variances were
521 high, and p values ranged from 0.12 to 0.75. Data are not shown for perseverations
522 given that there were no effects on any analysis.
523 There were group differences in Challenge Ratios for omissions and premature
524 responses for the noise and tone challenges. For the noise challenge, both the SHAM
19
525 and POR groups showed increases in the ratio of challenge omissions to baseline
526 omissions relative to the PER group (Figure 7A). This was evident in significant main
527 effects of group for SHAM vs. PER (F1,17=19.303, p=0.0001) and for POR vs. PER
528 (F1,14=5.827, p=0.03). In contrast, the PER showed increases in the ratio of challenge
529 premature to baseline premature responses relative to the both the SHAM and PER
530 groups. Again, this was evident in significant main effects of group for SHAM vs. PER
531 (F1,17=19.303, p=0.0001) and for POR vs. PER (F1,14=17.213, p=0.001). For the tone
532 challenge (Figure 7D), the PER also showed an increase in the ratio of the challenge
533 premature to baseline premature responses relative to the SHAM and PER groups.
534 Again, this was evident in significant main effects of group for SHAM vs. PER
535 (F1,17=19.303, p=0.0001) and for POR vs. PER (F1,14=17.213, p=0.001).
536 There were group differences in latency Challenge Ratios (Figure 8), but only for
537 the noise challenge. The pattern was similar for all three ratios in that the PER group
538 became faster with the noise challenge relative to baseline and the SHAM and POR
539 groups became slower (Figure 8A). ANOVA showed significant group differences for
540 PER vs. SHAM for correct latencies (F1,17=8.278, p=0.01) and marginally significant
541 differences for incorrect latencies (F1,17=3.724, p=0.07) and latency to retrieve the food
542 reward (F1,17=3.388, p=0.08). Differences were more robust for the PER vs. POR
543 comparison with significant group differences on Challenge Ratios for correct latencies
544 (F1,14=15.867, p=0.001), incorrect latencies (F1,17=3.724, p=0.07), and latencies to
545 retrieve the food reward (F1,14=6.776, p=0.021).
546
20
547 Discussion
548 In the present study, we used the 5CSRT task in combination with experimental
549 lesions to address the hypothesis that the PER and POR support different types of
550 attention. We compared rats with PER damage, POR damage, or SHAM control
551 surgeries on acquisition, training, and attentional challenges. The results a number of
552 revealed lesion effects among the PER, POR, and SHAM groups. During shaping, the
553 PER group showed impaired acquisition, differing from both the POR and SHAM groups
554 on multiple measures. During training and challenge sessions, they showed decreased
555 accuracies compared with POR and SHAM groups. During attentional challenges, the
556 PER rats were most affected by the presentation of noise right before target onset, and
557 the pattern of effects suggested increased impulsivity. During the noise challenge, the
558 PER group showed fewer omissions and more premature responses that either the POR
559 or SHAM groups. They also showed faster latencies to respond in the noise challenge,
560 whereas the POR and SHAM groups showed slower latencies to respond. The primary
561 impact of POR damage during shaping was facilitated acquisition compared with the
562 PER group. Numerically, the POR rats showed the fewest sessions to reach criterion,
563 and unlike PER and SHAM rats, none of the POR rats were advanced from one shaping
564 step to another without reaching criterion. During training, the POR group showed faster
565 latencies compared with PER and SHAM groups in the absence of speed accuracy
566 trade-off. This pattern suggests that the POR damage improves performance due to
567 reduced monitoring of the context allowing a more specific focus on the task
568 requirements resulting in increased attentional capacity, possibly due to better response
569 readiness and faster processing speed.
570 One contribution of the present study is that it addresses a controversy about
571 whether structures in the medial temporal lobe memory system are functionally
572 differentiated. Prior studies suggest that some medial temporal lobe structures may have
21
573 non-mnemonic functions, for example, attentional and perceptual functions (Bussey and
574 Saksida, 2007; Murray et al., 2007; Cordova et al., 2016). START The current findings
575 provide further evidence for the view that there is functional specialization in the medial
576 temporal lobe such that the function of some structures extends beyond the domains of
577 learning and memory.
578 One limitation of our study is that the differences in acquisition make it
579 challenging to interpret results of attentional challenges. Several PER animals and
580 some SHAM animals were advanced to the next limited hold without meeting
581 criterion during shaping, but no POR animals were advanced. The PER group
582 was clearly impaired in acquisition. The PER group was at about 67% accuracy
583 toward the end of shaping, and the POR and SHAM groups were at about 81%. For all
584 three groups, accuracy decreased by about 10% during training, possibly because we
585 dropped the stimulus duration too steeply during the end of shaping. However, the
586 profile was similar in that the POR and SHAM groups performed better than the PER
587 group. Moreover, accuracy during training for all three groups was stable for 20
588 sessions. For these reasons, we argue that results during the training phase and during
589 challenges provide useful and interpretable data.
590 During shaping, the PER group required more trials to reach criteria compared
591 with the SHAM group. The PER group also exhibited lower accuracies and higher
592 omissions during shaping. During training, the PER group was also significantly less
593 accurate than the SHAM and POR groups. This profile of performance was also evident
594 in baseline and challenge performances. The PER group showed fewer changes to
595 attentional challenges, perhaps because baseline performance was impaired and there
596 was less parametric space for impact of PER damage on variables. Interestingly, the
597 response of the PER group to the noise distraction challenge differed from that of the
22
598 SHAM and POR groups – the PER lesioned rats showed significantly increased
599 premature responses, generally interpreted as increased impulsivity. Since the PER
600 exhibits strong connectivity with the medial prefrontal cortex, PER damage may have
601 disrupted PER-medial prefrontal connections important for top-down control resulting in
602 increased impulsivity.
603 This pattern of results for the PER group suggests impaired learning, decreased
604 attentional capacity, and decreased impulse control. The impairments in learning and
605 accuracy at asymptote could represent perceptual impairments, but is more likely
606 attentional. The so-called perceptual deficits emerge in tasks in which animals are
607 required to disambiguate stimuli with complex overlapping features (Bussey and
608 Saksida, 2007; Kealy and Commins, 2011; Kent and Brown, 2012). We argue that these
609 deficits could be described as attentional deficits. In order to disambiguate stimuli with
610 overlapping features, animals must be able to select some features over others for
611 further processing. Most of the discrimination work addresses the visual domain, but
612 there is evidence for perirhinal involvement in all sensory domains including auditory
613 (Bang and Brown, 2009), olfactory (Herzog and Otto, 1998), gustatory (Tassoni et al.,
614 2000), and tactile (Holdstock et al., 2009) sensory domains. One possibility is that PER
615 damage impairs the animal’s ability to disambiguate nose poke holes in the 5CSRT
616 apparatus resulting in decreased accuracy and impaired acquisition. This could also be
617 extended to the increased sensitivity to the noise distraction in that rats with PER
618 damage may not have been able to disambiguate the onset of the noise from that of the
619 visual target. However, decreased impulse control seems to be the more likely
620 explanation for increased impulsivity when noise is presented prior to presentation of the
621 visual target.
622 During shaping, the POR group performed similarly to the SHAM group on all
623 measures except perseverations at the 64 s level for which the POR group showed on
23
624 average, about 2 more perseverations per session. Interestingly, on correct trials during
625 training the POR group exhibited significantly faster latencies to respond than either of
626 the other two groups. The faster latencies for the POR group were in the absence of a
627 decrease in accuracy, suggesting enhanced attention or greater response readiness for
628 the POR group. The POR group was also less impacted by attentional challenges
629 compared with the SHAM group, and the profile of responses differed particularly for the
630 noise challenge and for the variable ITI challenge. The SHAM group showed decreased
631 accuracies for the short and variable ITI challenges, but accuracies for the POR group
632 were not affected by any challenge. For the noise challenge both the POR and SHAM
633 groups showed increased omissions, but only the POR group showed longer latencies.
634 Longer latencies were evident for the variable ITI, although unlike the SHAM group, the
635 POR group did not show significantly decreased accuracy or increased omissions.
636 Overall, the main impact of challenges on the POR group was slower latencies to
637 respond with no change in accuracies, although response latencies were faster than
638 those of the SHAM group in all conditions.
639 We previously provided evidence for a postrhinal role in attention, specifically
640 attentional orienting (Bucci and Burwell, 2004). This is consistent with the anatomy in
641 that the POR is strongly and reciprocally connected with the lateral posterior nucleus of
642 the thalamus, which is the rodent homolog of the primate pulvinar and implicated in
643 attention (Agster et al., 2016; Tomas Pereira et al., 2016; Kaas and Baldwin, 2019; Yang
644 and Burwell, 2020). We have also provided evidence for the hypothesis that the POR
645 maintains a representation of the current context and monitors the context for changes,
646 updating the representation as changes occur (Burwell and Hafeman, 2003; Furtak et
647 al., 2012). One possibility is that, in the absence of an intact POR, rats were not
648 continuously monitoring the current environment and were able to focus more resources
649 on the task resulting in better response readiness and increased attentional capacity.
24
650 There is some indirect evidence for this interpretation. Earlier we showed that rats with
651 combined lesions of the PER and POR trained on a complex feature positive/feature
652 negative discrimination (FPFN) task showed facilitated acquisition (Gastelum et al.,
653 2012). We interpreted the facilitation as resulting from POR damage, because rats with
654 PER damage are impaired on the feature-negative task (LT1, T2) and on positive
655 patterning (LT1, L2, T2), when the compounds are presented simultaneously
656 (Campolattaro and Freeman, 2006b, a). Additionally, in the FPFN task, context is
657 thought to accrue inhibitory control over other cues. Without context representations this
658 inhibitory control would fail, and animals would be expected to learn the task more
659 efficiently. Accordingly, our interpretation is that the POR damage impaired contextual
660 control resulting in facilitated attentional monitoring. Thus, the present findings provide
661 more direct evidence for a POR role in stimulus driven, automatic attention.
662 How might PER and POR attentional functions be used? We have suggested
663 that the PER and POR support different types of attention in the service of learning and
664 memory. There is anatomical evidence for this assertion given that the PER and POR
665 provide direct inputs to the entorhinal cortex, subiculum, CA1, presubiculum, and
666 parasubiculum (Agster and Burwell, 2013). Indeed, an open question about episodic
667 memory is how items to be remembered are selected from all the possibilities available
668 at any given time. Regions outside the medial temporal lobe that connect with the PER
669 and POR/PHC have been proposed to serve this function (e.g., Cordova et al., 2016).
670 Here we provide evidence that the PER and POR/PHC, themselves, are important for
671 the selection of information to be remembered. It may also be the case that the PER and
672 POR support attentional and executive functions of other brain regions. For example, the
673 PER is strongly connected with medial prefrontal cortex, whereas the POR is strongly
674 connected with ventrolateral orbital prefrontal cortex. Based on functional and
675 anatomical evidence, we have suggested the PER and POR cortices act as a
25
676 contextual-support network that directly provides contextual and spatial information to
677 the prefrontal cortex (Peng and Burwell, under review). In addition to its robust
678 connections with the PER, the POR is also strongly interconnected with a number of
679 visuospatial processing regions including the posterior parietal cortex, retrosplenial
680 cortex, visual association cortex, and the pulvinar (Agster and Burwell, 2009, 2013;
681 Agster et al., 2016; Tomas Pereira et al., 2016; Yang et al., in press). Such connectivity
682 could support the POR binding of non-spatial information from the PER with spatial
683 information to represent the current local physical context, including the spatial layout of
684 objects and features in the environment as well as the geometry of the space. The POR
685 then automatically monitors the environment for changes and updates representations
686 when changes occur. These representations of context are available to be used by
687 multiple brain regions, including prefrontal, posterior cortical, and hippocampal areas, for
688 context-guided behavior, associative learning, and episodic memory (Estela-Pro and
689 Burwell, under review; Peng and Burwell, under revision).
690 In summary, the perirhinal (PER) and postrhinal (POR) cortices, structures in the
691 medial temporal lobe, are implicated in learning and memory. Whether medial temporal
692 lobe structures are exclusively dedicated to memory, however, is under debate. Here,
693 we provide evidence that the functions of the PER and POR cortices extend beyond
694 memory for objects and spatial memory for include attentional functions. Specifically, the
695 PER contributes to controlled attention, and the POR contributes to stimulus-driven
696 attention. These findings provide new evidence for functional specialization in the
697 medial temporal lobe.
698
699
700
701
26
702 References
703 Agster KL, Burwell RD (2009) Cortical efferents of the perirhinal, postrhinal, and 704 entorhinal cortices of the rat. Hippocampus 19:1159-1186.
705 Agster KL, Burwell RD (2013) Hippocampal and subicular efferents and afferents of the 706 perirhinal, postrhinal, and entorhinal cortices of the rat. Behav Brain Res.
707 Agster KL, Tomas Pereira I, Saddoris MP, Burwell RD (2016) Subcortical connections of 708 the perirhinal, postrhinal, and entorhinal cortices of the rat. II. efferents. Hippocampus 709 26:1213-1230.
710 Ahn JR, Lee I (2017) Neural Correlates of Both Perception and Memory for Objects in 711 the Rodent Perirhinal Cortex. Cereb Cortex 27:3856-3868.
712 Bang SJ, Brown TH (2009) Perirhinal cortex supports acquired fear of auditory objects. 713 Neurobiol Learn Mem 92:53-62.
714 Bari A, Dalley JW, Robbins TW (2008) The application of the 5-choice serial reaction 715 time task for the assessment of visual attentional processes and impulse control in rats. 716 Nat Protoc 3:759-767.
717 Beaudin SA, Singh T, Agster KL, Burwell RD (2013) Borders and comparative 718 cytoarchitecture of the perirhinal and postrhinal cortices in an F1 hybrid mouse. Cereb 719 Cortex 23:460-476.
720 Bucci DJ, Burwell RD (2004) Deficits in attentional orienting following damage to the 721 perirhinal or postrhinal cortices. Behavioral Neuroscience 118:1117–1122.
722 Burwell RD, Hafeman DM (2003) Positional firing properties of postrhinal cortex neurons. 723 Neuroscience 119:577-588.
724 Burwell RD, Witter MP, Amaral DG (1995) Perirhinal and postrhinal cortices of the rat: a 725 review of the neuroanatomical literature and comparison with findings from the monkey 726 brain. Hippocampus 5:390-408.
727 Bussey TJ, Saksida LM (2007) Memory, perception, and the ventral visual-perirhinal- 728 hippocampal stream: thinking outside of the boxes. Hippocampus 17:898-908.
729 Bussey TJ, Muir JL, Aggleton JP (1999) Functionally dissociating aspects of event 730 memory: the effects of combined perirhinal and postrhinal cortex lesions on object and 731 place memory in the rat. J Neurosci 19:495-502.
732 Campolattaro MM, Freeman JH (2006a) Perirhinal cortex lesions impair feature-negative 733 discrimination. Neurobiol Learn Mem 86:205-213.
734 Campolattaro MM, Freeman JH (2006b) Perirhinal cortex lesions impair simultaneous 735 but not serial feature-positive discrimination learning. Behav Neurosci 120:970-975.
736 Carli M, Robbins TW, Evenden JL, Everitt BJ (1983) Effects of lesions to ascending 737 noradrenergic neurones on performance of a 5-choice serial reaction task in rats;
27
738 implications for theories of dorsal noradrenergic bundle function based on selective 739 attention and arousal. Behav Brain Res 9:361-380.
740 Cordova NI, Tompary A, Turk-Browne NB (2016) Attentional modulation of background 741 connectivity between ventral visual cortex and the medial temporal lobe. Neurobiol Learn 742 Mem 134 Pt A:115-122.
743 Cowell RA, Bussey TJ, Saksida LM (2010) Components of recognition memory: 744 dissociable cognitive processes or just differences in representational complexity? 745 Hippocampus 20:1245-1262.
746 Eacott MJ, Gaffan EA (2005) The roles of perirhinal cortex, postrhinal cortex, and the 747 fornix in memory for objects, contexts, and events in the rat. Q J Exp Psychol B 58:202- 748 217.
749 Eichenbaum H, Yonelinas AP, Ranganath C (2007) The medial temporal lobe and 750 recognition memory. Annu Rev Neurosci 30:123-152.
751 Estela-Pro VJ, Burwell RD (under review) The Anatomy and Function of the Postrhinal 752 Cortex. Behavioral Neuroscience.
753 Furtak SC, Ahmed OJ, Burwell RD (2012) Single neuron activity and theta modulation in 754 postrhinal cortex during visual object discrimination. Neuron 76:976-988.
755 Gastelum ED, Guilhardi P, Burwell RD (2012) The effects of combined perirhinal and 756 postrhinal damage on complex discrimination tasks. Hippocampus 22:2059-2067.
757 Gaynor LS, Johnson SA, Mizell JM, Campos KT, Maurer AP, Bauer RM, Burke SN 758 (2018) Impaired discrimination with intact crossmodal association in aged rats: A 759 dissociation of perirhinal cortical-dependent behaviors. Behav Neurosci 132:138-151.
760 Heimer-McGinn VR, Poeta DL, Aghi K, Udawatta M, Burwell RD (2017) Disconnection of 761 the Perirhinal and Postrhinal Cortices Impairs Recognition of Objects in Context But Not 762 Contextual Fear Conditioning. J Neurosci 37:4819-4829.
763 Herzog C, Otto T (1998) Contributions of anterior perirhinal cortex to olfactory and 764 contextual fear conditioning. NeuroReport 9:1855-1859.
765 Holdstock JS, Hocking J, Notley P, Devlin JT, Price CJ (2009) Integrating visual and 766 tactile information in the perirhinal cortex. Cereb Cortex 19:2993-3000.
767 Jarrard LE, Davidson TL, Bowring B (2004) Functional differentiation within the medial 768 temporal lobe in the rat. Hippocampus 14:434-449.
769 Kaas JH, Baldwin MKL (2019) The Evolution of the Pulvinar Complex in Primates and Its 770 Role in the Dorsal and Ventral Streams of Cortical Processing. Vision (Basel) 4.
771 Kealy J, Commins S (2011) The rat perirhinal cortex: A review of anatomy, physiology, 772 plasticity, and function. Prog Neurobiol 93:522-548.
28
773 Kent BA, Brown TH (2012) Dual functions of perirhinal cortex in fear conditioning. 774 Hippocampus 22:2068-2079.
775 Lawrence AV, Cardoza J, Ryan L (2020) Medial temporal lobe regions mediate complex 776 visual discriminations for both objects and scenes: A process-based view. Hippocampus 777 30:879-891.
778 Maurer AP, Nadel L (2021) The Continuity of Context: A Role for the Hippocampus. 779 Trends Cogn Sci 25:187-199.
780 Murray EA, Bussey TJ, Saksida LM (2007) Visual Perception and Memory: A New View 781 of Medial Temporal Lobe Function in Primates and Rodents. Annu Rev Neurosci.
782 Peng X, Burwell RD (under review) Beyond the hippocampus: the role of 783 parahippocampal-prefrontal communication in context-modulated behavior. 784 Neurobiology of Learning and Memory.
785 Peng X, Burwell RD (under revision) Beyond the hippocampus: the role of 786 parahippocampal-prefrontal communication in context-modulated behavior. 787 Neurobiology of Learning and Memory.
788 Ramos JM (2013) Differential contribution of hippocampus, perirhinal cortex and 789 postrhinal cortex to allocentric spatial memory in the radial maze. Behav Brain Res 790 247:59-64.
791 Squire LR, Stark CE, Clark RE (2004) The medial temporal lobe. Annu Rev Neurosci 792 27:279-306.
793 Tassoni G, Lorenzini CA, Baldi E, Sacchetti B, Bucherelli C (2000) Role of the perirhinal 794 cortex in rats' conditioned taste aversion response memorization. Behav Neurosci 795 114:875-881.
796 Tomas Pereira I, Agster KL, Burwell RD (2016) Subcortical connections of the perirhinal, 797 postrhinal, and entorhinal cortices of the rat. I. afferents. Hippocampus 26:1189-1212.
798 Winstanley CA, Chudasama Y, Dalley JW, Theobald DE, Glennon JC, Robbins TW 799 (2003) Intra-prefrontal 8-OH-DPAT and M100907 improve visuospatial attention and 800 decrease impulsivity on the five-choice serial reaction time task in rats. 801 Psychopharmacology (Berl) 167:304-314.
802 Yang FC, Burwell RD (2020) Neuronal Activity in the Rat Pulvinar Correlates with 803 Multiple Higher-Order Cognitive Functions. Vision (Basel) 4.
804 Yang FC, Dokovna LB, Burwell RD (in press) Cerebral Cortex.
805
806
29
807 Tables
808
809
810 Table 1: Lesion Coordinates Region AP ML DV PER -3.3 5.7 -6.7 -4.3 5.7 -6.7 -5.3 5.7 -6.7 -6.3 5.7 -6.7 -7.4 5.7 -6.2 POR 0.54 5.3 -5.5 -0.46 5.3 -5.5 -1.46 5.3 -5.5 811 Perirhinal anterior-posterior (AP) coordinates are 812 measured in mm from Bregma. Postrhinal 813 anteroposterior (AP) coordinates are measured in mm 814 from Lambda. Mediolateral (ML) coordinates are 815 measured in mm to the left and to the right from the 816 midline. Dorsoventral (DV) coordinates are measured in 817 mm from skull. 818
819
820
821
30
822 Table 2: Response to Challenges Region Noise prior to Short Variable Tone Target Target ITI during ITI SHAM Accuracy - ↓↓↓ ↓↓↓ - Premature Responses - - - - Omissions ↑↑ - ↑↑↑ - Perseverations - - - - Correct Latencies - - ↑↑ - Incorrect Latencies - - ↑↑↑ - Reward Latencies - - ↓↓ - PER Accuracy - - - - Premature Responses ↑↑↑ - ↓↓ ↑ Omissions - - - - Perseverations - - - - Correct Latencies ↓ - ↑ - Incorrect Latencies ↓ - ↑↑ - Reward Latencies - - - - POR Accuracy - - - - Premature Responses - - - - Omissions ↑↑ ↑ - - Perseverations - - - - Correct Latencies ↑↑ - ↑↑↑ - Incorrect Latencies ↑↑ - ↑↑↑ - Reward Latencies ↑ - ↓↓ - 823 Results of within subject repeated measures ANOVA in which each attentional challenge was 824 compared with its own baseline for each lesion group. Symbols: ↑↑↑, p<0.001; ↑↑, 825 p<0.05; ↑, p<0.075. Abbreviations: SHAM = controls, PER = perirhinal cortex, POR = 826 postrhinal cortex. 827 828 829 830 831 832 833
31
834 Figure Legends 835 836 Figure 1. Coronal sections showing the extent of experimental lesions. Schematics of 837 largest and smallest lesions are shown for the PER group (A) and the POR group (B). 838 The largest lesion is shown in dark grey and the smallest lesion is shown in light grey. 839 Scale bar = 1 mm. 840 841 Figure 2. Examples of experimental lesion damage in the PER and the POR. Damage 842 was identified as missing cortex, thinning cortex, missing cells, and pyknotic cells. A. 843 In this example from a PER subject, there was damage to the superficial layers and 844 middle layers of the dorsoventral extent of the PER. Layer I is thinner than in control 845 subjects. There are areas in which cells have disappeared or have become pyknotic in 846 superficial layers and deep layer V. B. This POR subject shows damage to deep layers 847 exhibited as pyknotic cells as well as thinning of cortical layers. Again, secondary 848 damage would be expected in superficial layers due to the death of deep layer cells 849 that project to superficial layers. In both cases, there is likely secondary damage that is 850 not apparent. See text for details. In this case, there is some damage in dorsally 851 adjacent temporal ventral cortex (TEv) and possibly to the ventrally adjacent lateral 852 entorhinal area (LEA). In both cases, there is likely secondary damage that is not 853 apparent. See text for details. Scale bar = 500 µm. 854 855 Figure 3. Group performance during shaping. A. The PER group compared with both 856 the SHAM and the POR groups required significantly more sessions to reach criterion. 857 They also showed lower accuracies (B), higher omissions, especially in later shaping 858 stages (C), and longer latencies to make an incorrect choice, especially early in 859 shaping (D). E. The POR group also showed marginally higher premature responses 860 than the PER group. F. The POR group differed from the SHAM group only in the 861 pattern of perseverations, showing more perseverations at some limited holds and 862 fewer at others. No other differences in shaping were observed. Abbreviations: InS, 863 initial shaping. Group differences: * = p<0.05, ** = p<0.01, *** = p<0.001, # = p<0.075. 864 POR vs. SHAM, Group by Limited Hold: p<0.0001, xxx. 865 866 Figure 4. Group performance during training. A. Percent accuracy across blocks of five 867 sessions. B. Overall percent accuracy. C. Latencies to respond on correct trails across 868 blocks of five sessions. D. Overall latencies on correct trials. The PER group exhibited 869 significantly lower accuracies than the POR group and marginally significantly lower 870 accuracy than the SHAM group. The POR group exhibited significantly faster latencies 871 on correct training trials than the PER group and marginally significantly faster 872 latencies than the SHAM group. Abbreviations: InS, initial shaping. Group differences: * 873 = p<0.05, # = p<0.075. 874 875 Figure 5. Performance during baseline and attentional challenges. Shown in each 876 panel are the mean of the challenge baseline sessions (Ave BL) along with mean 877 challenge performance along with group differences. Ave BL are shown for simplicity 878 though each challenge was compared to its own baseline. It is important to note that, 879 because there were differences in acquisition, baselines differ across groups. A. For 880 percent (Pct) accuracy, the PER group was significantly or marginally significantly 881 lower than both SHAM and POR groups on Ave BL and each of the challenges. B. For 882 Pct Omissions, the PER group was marginally significantly higher than SHAM and 883 POR groups on baseline performance, and significantly higher than both on Short 884 challenges. The PER group was also significantly higher than the POR group n
32
885 variable ITI and Tone challenges. C. For Pct Premature responses, the ER group was 886 significantly higher than the SHAM group on Noise challenges. Group differences: * = 887 p<0.05, ** = p<0.01, *** = p<0.001, # = p<0.075. Solid lines indicate main effects of 888 group and dashed lines indicate significant group by condition interactions. 889 890 Figure 6. Latencies to respond during baseline and attentional challenges. Shown in 891 each panel are the mean average baselines (Ave BL) latencies and challenge latencies 892 along with group differences. A. For correct trials, the POR group was significantly or 893 marginally significantly faster than both SHAM and PER groups on baseline and 894 challenges. For the noise challenge, there were group by condition interactions such 895 that the PER group was faster relative to baseline than the POR or SHAM groups. B. 896 For latencies on incorrect trials, the POR group was significantly faster than SHAM and 897 PER on baseline performance and the tone challenge. C. For latencies to retrieve 898 rewards there was a marginally significant group by condition interaction between the 899 PER and POR groups for the tone challenge. The POR group was faster at baseline 900 and the tone challenge, but the difference was less for the challenge. Significance: * = 901 p<0.05, ** = p<0.01, *** = p<0.001, # = p<0.075. Solid lines indicate main effects of 902 group and dashed lines indicate significant group by condition interactions. 903 904 Figure 7. Impact of attentional challenges on accuracy, omissions, and premature 905 responses during attentional challenges. Shown are ratios of performance on 906 attentional challenges relative to baseline for accuracy, omitted trials, and premature 907 responses. Differences here reflect group by condition interactions (prior baseline vs 908 challenge). The Challenge Ratio is calculated as CR = (Challenge- 909 Baseline)/(Challenge+Baseline) such that scores near zero indicate little change from 910 baseline during the challenge. Positive and negative scores indicate increases and 911 decreases from baseline, respectively. R is shown for the noise distraction during 912 target presentation (A). Both the SHAM and POR groups showed significantly 913 increased omissions relative to baseline. There were no significant group by condition 914 interactions for shortened target presentations (B), variable ITIs (C), or the tone 915 presented during the ITI (D). Both the SHAM and POR groups showed more omissions 916 during the Noise challenge (A) Group differences: * = p<0.05, *** = p<0.001, # = 917 p<0.075. 918 919 Figure 8. Impact of attentional challenges on latencies to respond. Shown are ratios of 920 performance on attentional challenges relative to baseline for latencies on correct and 921 incorrect trials as well as latencies to retrieve rewards. As in Figure 7, differences here 922 reflect group by condition interactions. The Challenge Ratio is calculated as CR = 923 (Challenge-Baseline)/(Challenge+Baseline) such that scores near zero indicate little 924 change from baseline during the challenge. Positive and negative scores indicate 925 increases and decreases from baseline, respectively. (A). For both the SHAM and 926 POR groups latencies changed in directions opposite that of the PER group relative to 927 baseline. There were no significant group by condition interactions for shortened target 928 presentations (B), variable ITIs (C), or the tone presented during the ITI (D). Both the 929 SHAM and POR groups showed more omissions during the Noise challenge (A) 930 Group differences: * = p<0.05. 931 932
33 A
36 36 35 35
-3.80 mm -4.80 mm
36 36 35
-6.05 mm -7.00 mm
B POR
POR
-7.80 mm -8.72 mm AB TEv TEv
POR
PER
LEA
LEA A 20 SHAM D 30 18 PER 16 25 14 POR 20 * * 13 10 15
Sessions 8 * * * 6 10 4 5 2 0 LatenciesIncorrect (s) 0 InS 32 16 8 4 2 1.5 1.0 64 32 16 8 4 2 1.5 1.0 Limited Hold (s) Limited Hold (s) B 100 E 14 12 80 * * * 10 # 60 8
40 6 4 20 Percent Accuracy Percent 2 PrematureResponses 0 0 64 32 16 8 4 2 1.5 1.0 64 32 16 8 4 2 1.5 1.0 Limited Hold (s) Limited Hold (s) C 40 F 10
8 30 6 * * 20 * 4 xxx 10
Perseverations 2 PercentOmissions 0 0 64 32 16 8 4 2 1.5 1.0 64 32 16 8 4 2 1.5 1.0 Limited Hold (s) Limited Hold (s) A B SHAM 80 80 # * 70 70 PER 60 60 POR 50 50 40 40 SHAM 30 30 20 PER 20 Percent Accuracy Percent Percent Accuracy Percent 10 POR 10 0 0 1 2 3 4 5 Training Blocks of 4 Sessions Accuracy C D # 1.2 1.2 * 1.0 1.0 0.8 0.8 0.6 0.6 Latency(s) Latency(s) 0.4 0.4 0.2 0.2
0.0 0.0 1 2 3 4 5 Correct Training Blocks of 4 Sessions Latencies ***p<0.001 **p<0.01 C B A *p<0.05 # p<0.075 Pct Premature Pct Omissions Pct Accuracy 10 15 20 25 35 30 40 45 35 10 15 20 25 10 20 30 40 50. 60 70 80 5 0 5 0 0 v L os hr a T Tone ITI var Short Noise Ave BL v L os hr a T Tone ITI var Short Noise Ave BL v L os hr a T Tone ITI var Short Noise Ave BL * *** # Effects are maineffects ofgroup *** * # * * # *** * * *** * POR PER SHAM ** *** * # * *** * * # A 1.5 # * ** * * * ** * * 1.0
0.5 CorrLatency (sec) 0 Ave BL Noise Short var ITI Tone B 3.5 * SHAM PER POR 3.0 ** * * 2.5 * 2.0 1.5 1.0 0.5
IncorrLatency (sec) 0 Ave BL Noise Short var ITI Tone C 1.5 *
1.0
0.5
0 RewardLatency (sec) Ave BL Noise Short var ITI Tone 0.7 *** *** 0.6 A SHAM * * B 0.6 0.5 PER 0.5 0.4 0.4 POR 0.3 0.3 0.2 0.2 0.1 0.1 0 0 -0.1 NoiseCR -0.2 CR Short -0.1 -0.3 -0.2 -0.4 -0.3 -0.5 -0.4 Accuracy Omissions Premature Accuracy Omissions Premature C 0.5 D 0.5 * 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 -0.1 -0.1 -0.2 CR Tone -0.2 Variable CR Variable -0.3 -0.3 -0.4 -0.4 -0.5 -0.5 Accuracy Omissions Premature Accuracy Omissions Premature C A Variable CR Noise CR -0.3 -0.2 -0.1 -0.1 0.1 0.2 0.1 0.2 0.3 0 0 Correct Incorrect Reward CorrectIncorrectReward * *** # *** POR PER SHAM * B D Tone CR Short CR -0.2 -0.1 -0.2 -0.1 0.1 0.1 0 0 Correct Incorrect Reward CorrectIncorrectReward