- Email: [email protected]
Short-term memory for food reward magnitude: The role of the prefrontal cortex
Behavioural Brain Research 88 (1997) 239 – 249
Research report
Short-term memory for food reward magnitude: The role of the prefrontal cortex William E. DeCoteau a, Raymond P. Kesner a,*, Joseph M. Williams b a
Department of Psychology, Uni6ersity of Utah, Salt Lake city, UT 84112, USA b Department of Psychology, Ohio State Uni6ersity, Columbus, OH, USA
Received 30 October 1996; received in revised form 26 February 1997; accepted 27 February 1997
Abstract Memory for magnitude of reinforcement was assessed in rats using a go/no-go short-term memory paradigm. During the task’s study phase rats were given a piece of cereal comprised of either 25 or 50% sugar. For all trials, one of the cereal types was designated positive, the other negative. On the ensuing test phase the rat was presented with an object which covered a food well. If a positive food reward was given during the study phase, a second food reward was placed beneath the object. No food reward was placed under the object if the study phase consisted of a negative food reward. Latency to object displacement was used as the measure of performance. Following the establishment of a significant difference between latency to approach the object with reward compared to latency to approach the object without reward, rats were given either agranular insular cortex, anterior cingulate cortex, pre- and infralimbic cortex or control lesions. Agranular insular cortex lesioned animals demonstrated a mild post-surgery impairment. Trials consisting of 10 and 20 s delays between the study and test phases were then introduced. Delays exacerbated the previous deficit of the agranular insular cortex lesion group, but had little effect on the other lesion groups. All animals transferred to a new set of cereals containing 25 and 50% sugar and exhibited taste preferences to sugar solutions of different concentrations. These results indicate that the agranular insular cortex may play an important role in the processing of affect-laden information within a prefrontal cortex short-term or working memory system. © 1997 Elsevier Science B.V. Keywords: Prefrontal cortex; Agranular insular cortex; Working memory; Reward; Rat
1. Introduction Kesner has proposed that new information is represented within a data-based memory system by multiple forms or attributes of memory [10]. In support of this proposal it has been demonstrated that in tasks that require novel and/or unique information for each trial, the hippocampus mediates short-term memory for spatial and temporal attributes, the extrastriate and perirhinal cortex mediate short-term memory for visual object information as an example of a sensory-per* Corresponding author. Tel.: + 1 801 581 7430; fax: +1 801 581 5841; e-mail: [email protected] 0166-4328/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S 0 1 6 6 - 4 3 2 8 ( 9 7 ) 0 0 0 4 4 - 2
ceptual attribute, the caudate-putamen mediates shortterm memory for a response attribute and amygdala mediates short-term memory for an affect attribute [9,12,19,24]. In recent years it has become clear that there is not just one neural region, but rather a neural circuit, that mediates short-term memory for specific attributes within the data-based memory system. For example, for the spatial attribute short-term memory representations can be found in the pre- and parasubiculum, medial entorhinal cortex and pre- and infralimbic cortex in addition to the hippocampus [8] (and unpublished observations). Similarly, it is highly likely that a neural circuit mediates the affect attribute as measured by memory for magnitude of reinforce-
240
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
ment. To test this idea the present study is focused on the agranular insular cortex. This structure was chosen because it has been shown to support intracranial selfstimulation [26], lesions to the area, when combined with adjacent gustatory cortex, produce impairments in taste aversion learning [2,13,17] and taste-potentiated odor aversion [16], and because of its direct neural connections with the amygdala [1,6,21,27]. Thus, the first purpose of this study is to examine the role of the agranular insular cortex in a successive delayed conditional discrimination task for magnitude of reward value. In the study phase of the task, rats were given one of two cereals. One cereal contained 25% sugar; the other 50% sugar. One of the two cereals was always designated as the positive stimulus and the other as the negative stimulus. This study phase was followed by the test phase in which the rat was shown an object which covered a food well. If the rat was given the positive food stimulus during the study phase, another food reward was placed beneath the object. Rats learn to approach the objects quickly when they expect a reward and they are slow to approach the object when they expect no reward. Recent research using this task has demonstrated that rats with amygdala lesions are impaired in short-term memory for reward value, even though they display normal preference for differential reward [12]. In previous work, it has been shown that short-term or working memory for different forms or attributes of memory is distributed across different neural regions within the prefrontal cortex of the rat. For example, pre- and infralimbic cortex mediates short-term memory for visual object and spatial attributes, whereas the anterior cingulate cortex and medial precentral cortex mediate the response attribute [11]. In order to determine whether short-term memory for the affect attribute is uniquely subserved by the agranular insular cortex, the second purpose of this study is to examine pre- and infralimbic cortex and the anterior cingulate cortex to ascertain the role of these prefrontal structures in mediating the affect attribute. The pre- and infralimbic cortical area is of particular interest since, like the agranular insular cortex, it has direct reciprocal connections with the amygdala [27,28].
2. Materials and methods
2.1. Subjects Nineteen male Hooded Long-Evans rats, weighing 300 – 350 g at the start of the experiment, were used as subjects. They were housed in standard stainless steel cages in a large, well lit laboratory room and were maintained on a 12:12 h light:dark schedule. All animals were placed on food deprivation with ad lib water
and maintained at 80–85% of ad lib weight throughout the experiment.
2.2. Apparatus The apparatus consisted of a platform, 122.0 cm long by 35.5 cm wide, with a black Plexiglas door (50.5 cm tall by 43.0 cm wide) separating the table into two halves. A pulley system enabled the experimenter to raise or lower the black Plexiglas door. The table was constructed of painted black wood and was raised 91.0 cm above the floor. Each side contained three 2.6 cm diameter food wells, separated by 6.0 cm, located 5.0 cm from the back edge, and centered from side-to-side. A sheet of red Plexiglas (88.0 cm tall by 91.0 cm wide) extended across one side of the table to block the animal’s view of the experimenter and the room; the opposite side of the table was against a wall. The stimuli for this experiment were a variety of three dimensional objects.
2.3. Procedure Rats were given a 3 week pre-training period. Rats initially learned to obtain food from the food wells. They were subsequently conditioned to push aside a 3-dimensional object covering the food well in order to obtain a food reward. During this pre-training phase, rats were habituated to three cereals: Oat Bake which contains 25% sugar; Froot Loops which contains 50% sugar and Apple Jacks (50% sugar) which was to be used as a neutral food stimulus. Once the rats learned to uncover the food wells, the training sessions began. The rats always began each trial at the same end of the apparatus. At the start of a trial (the study phase) the black Plexiglas door was raised allowing the rat to enter the other side of the apparatus. The door was lowered after the rat crossed to the other side. In the study phase of the trial, the rat encountered a 3-dimensional object covering the center food well. The objects used for each rat remained constant throughout the experiment. Located in the food well was one of two cereals: either an Oat Bake or a Froot Loop. One of the two cereals was always designated as the positive stimulus and the other was always designated as the negative stimulus. This assignment was chosen randomly but remained constant throughout the testing period for each rat. Each rat received ten trials per day, 5 days per week. Five trials were designated as positive and five trials were designated as negative. After the rat consumed the cereal, the opaque Plexiglas door was again raised allowing the rat access to the other side of the apparatus. This constituted the start of the test phase. During the test phase, the rat encountered a different object covering the center food well. If
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
the rat received the positive stimulus cereal during the study phase, in the subsequent test phase, another food reward (the neutral cereal) was placed beneath the object. If the rat received the negative stimulus during the study phase, no food reward was placed in the center food well during the subsequent test phase. Latency measured from crossing the center to the displacement of the object to uncover the center food well during the test phase was used as the dependent measure. The trial ended when the rat displaced the object from over the foodwell or ten seconds had elapsed.
2.4. Surgery After rats reached a criterion level of at least a 4.5 s mean latency difference between the positive and negative trials for two consecutive blocks of 50 trials, they were given either agranular insular cortex, pre- and infralimbic cortex, anterior cingulate cortex, or sham operated lesions. There were seven rats in the agranular insular cortex group, and four rats each in the pre- and infralimbic cortex, anterior cingulate cortex, and sham control groups. All surgeries were performed under sodium pentobarbital anesthesia (50 mg/kg, ip). All lesion sites were based on the Paxinos and Watson [23] rat stereotaxic atlas. Lesions to the agranular insular cortex, and pre- and infralimbic cortex were made electrolytically using stainless steel electrodes that were insulated except for a 0.5 mm tip. Direct current (1.2 mA) was applied for 15 s for each agranular insular cortex electrode placement and for 10 s for each preand infralimbic cortex placement. The coordinates for the agranular insular cortex lesion were as follows: 4.5 mm posterior to bregma, 3.2 mm lateral to midline, 2.4 mm ventral to dura; 3.5 mm posterior to bregma, 4.3 mm lateral to midline, 3.8 mm ventral to dura; and 2.5 mm posterior to bregma, 5.0 mm lateral to midline, 3.8 mm ventral to dura. Pre- and infralimbic lesions were made using the following coordinates: 3.2 mm posterior to bregma, 0.5 mm lateral to midline, 2.5 and 4.2 mm ventral to dura; and 2.2 mm posterior to bregma, 0.5 mm lateral to midline, 3.1 mm ventral to dura. Anterior cingulate cortex lesioned rats had the bone and dura removed above the lesion site. The region 0.5 mm anterior to bregma to the frontal pole, 1.0 lateral to midline, and 3.0 mm ventral to dura was then aspirated. All rats were then provided with a 1 week recuperative period. Post-surgery testing involved 200 trials on the same task learned prior to surgery. Following these 200 post-surgery trials, rats were tested at delays of 10 and 20 s between the study phase and the test phase. Rats received 100 trials each at 10 s and at 20 s delays. Subsequent to the 20 s delay testing, rats were returned to the no delay condition and given transfer tests to assess whether rats were using odor, texture
241
and/or visual cues to solve the task rather than sweetness differences between the two cereals. Rats received the same ten trials per day as before. However, for two of the ten trials, a novel set of cereals was substituted. The new cereals contained approximately the same percentage of sugar as the original cereals. Almond Delight, which contains approximately 25% sugar, was used as a substitute for Oat Bake. Kellogg’s Honey Smacks, which contain approximately 50% sugar, was used as a substitute for Froot Loops. As a final control, rats were given a preference task to determine whether impairments on the original task were the result of difficulty with memory for sweetness intensity rather than an inability to discriminate between sweetness intensity. In the preference task, the rats were allowed 15 min of water per day and an unlimited supply of food. Prior to testing, rats were habituated to 15 min of water per day. Following habituation, rats alternated between days where they were presented with two bottles, one containing a 25%, the other a 50% sugar solution, and days where they were given two bottles containing only water. The water condition served as a control for the sugar solution condition. To prevent any side bias the placement of each sugar solution bottle was randomly located to the left or right during each day of testing. The observer recorded the amount of water the rat drank from each bottle. Rats were tested on this task for eight consecutive days.
2.5. Histological 6erification Following these tests, rats were sacrificed with an overdose of sodium pentobarbital (50 mg/kg, ip) and perfused with 10% formalin. The brains were removed and stored in 30% sucrose formalin for 1 week. They were then frozen for slide preparation. Transverse sections (24 mm) were cut with a cryostat through the lesion area and stained with cresyl violet. Brain sections were examined for histological accuracy of the lesion placement.
3. Results
3.1. Histology Histological analysis of the agranular insular cortex lesioned brains revealed that two animals had damage immediately dorsal to the intended site and were thus excluded from the agranular insular cortex group and added to the sham control group. For these two animals the anterior portion of their lesions consisted of bilateral destruction to the lateral frontal cortex, area 2, and parietal cortex, area 1, and the posterior aspect of their lesions damaged granular and dysgranular insular
242
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
Fig. 1. Illustrations of the smallest (black) and largest (stippled) agranular insular cortex lesions.
cortex. The overall degree of cortical tissue damage in these animals was very similar to the amount lost in the animals included in the agranular insular cortex lesioned group. Representative examples of the smallest
and largest of the five remaining agranular insular cortex lesions are shown in Fig. 1. Examination of these animals indicated that they all received bilateral removal along the entire anterior-posterior axis of the
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
243
Fig. 2. Illustrations of the smallest (black) and largest (stippled) pre- and infralimbic cortex lesions.
agranular insular cortex. The anterior components of the larger lesions contained some damage to the ventral lateral portions of frontal cortex, areas 2 and 3. The posterior aspect of the larger lesions included damage to ventral parietal cortex, area 1, and to the dysgranular insular cortex. Granular insular cortex was consistently spared in all agranular insular cortex lesions.
Figs. 2 and 3 show representative examples of the smallest and largest pre- and infralimbic and anterior cingulate lesions, respectively. Histological analysis of the pre- and infralimbic cortex lesioned brains revealed bilateral destruction of the pre- and infralimbic cortex. Larger lesions included additional damage to the posterior aspect of ventromedial orbital cortex, cingulate cortex, area 3 and the medial aspect of the caudate-
244
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
Fig. 3. Illustrations of the smallest (black) and largest (stippled) anterior cingulate cortex lesions.
putamen. All anterior cingulate cortex lesions included damage to frontal cortex, area 2 and cingulate cortex, area 1. The anterior portion of the lesion included removal of cingulate cortex, area 3.
3.2. Memory for magnitude of reinforcement Analyses of the two animals excluded on histological grounds from the agranular insular cortex group indi-
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
245
Fig. 4. Pre-and post-operative mean latency for positive and negative trials for control. (a), agranular insular cortex (b), pre- and infralimbic cortex (c) and anterior cingulate cortex (d) lesioned animals.
cated that they were not significantly different from the sham control group in either pre- or post-surgery performance. The animals were thus combined to form one control group (n=6) for the remainder of the analyses. The mean number of trials to learn this task was 475 for the control group, 480 for the agranular insular cortex group, 513 for the pre- and infralimbic cortex group and 538 for the anterior cingulate cortex group. A one-way analysis of variance (ANOVA) indicated that the differences among the four groups were not significant. The effects of the four types of lesions on pre- and post-operative mean latency for positive and negative trials are shown in Fig. 4 (a,b,c and d). The data were grouped into blocks of 100 trials for analysis. This included the criterion trials (Pre) and 2 sets of post surgery trials (Post 1 and Post 2). A repeated measures
three-way ANOVA with lesion groups as the between factor and pre-post 1-post 2 (blocks) and positive-negative trial type as the within factors revealed that there was a significant lesion effect, F3,15 = 10.3, PB0.001, a significant block effect, F2,30 = 3.7, PB 0.05, a significant trial type effect, F1,15 = 545.0, PB0.0001, a significant block by lesion interaction effect, F6,30 =3.1, PB 0.01, a significant trial type by lesion interaction effect, F3,15 = 4.4, PB 0.05, and a significant block by trial type interaction effect, F2,30 = 4.6, PB .01. Further Newman-Keuls comparison tests of the block by lesion interaction indicated that there were no significant differences among the four groups with respect to pre-operative performance and the pre- and infralimbic cortex, anterior cingulate cortex and cortical control animals showed no significant deficit in their post-operative compared to pre-operative performance. The
246
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
Fig. 5. Mean latencies during 1–4, 10, and 20 s delay conditions for positive and negative trials for control (a), agranular insular cortex (b), preand infralimbic cortex (c) and anterior cingulate cortex (d) lesioned animals.
agranular insular cortex lesioned group, however, had a deficit in performance in both the first and second post-operative blocks compared to pre-operative performance (P B 0.05). Newman-Keuls comparison tests of the trial type by lesion interaction revealed that the impairment of the agranular insular cortex group was due to poor performance on negative trials (P B0.05). The effects of agranular insular cortex, pre- and infralimbic cortex, anterior cingulate cortex or control lesions on mean latency performance for the positive and negative trials at 1 – 4, 10, and 20 s delay conditions are shown in Fig. 5 (a – d). Data were again grouped into blocks of 100 trials for analysis which included the last 100 trials of the immediate (1 – 4 s) condition and 100 trials each for the 10 and 20 sec conditions. A repeated measures three-way ANOVA with lesion groups as the between factor and delays and trial type as the within factors revealed that there was a signifi-
cant lesion effect F3,15 = 20.2, PB 0.0001, a significant block effect F2,30 = 7.6, PB 0.005, a significant trial type effect F1,15 = 237.6, PB0.0001, a significant block by trial type interaction effect F2,30 = 6.4, PB0.005, and a significant trial type by lesion interaction effect F3,15 = 8.9, P B0.001. Further Newman-Keuls comparison tests of the trial type by lesion interaction indicated that the agranular insular cortex group was impaired compared to the other three lesion groups and that this deficit was due to poor performance on negative trials (PB 0.05). It is important to note that, in observing the behavior of the animals executing the task, it was clear that the poor performance of the agranular insular lesioned animals during both the no delay and delay conditions was not due to a change in running speed. Immediately upon the lifting of the guillotine door the animals of all lesion groups would quickly run the length of appara-
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
247
tus. It was not until they were within its close proximity that the rats decided whether or not to displace the test phase object. At this moment impairments became observable. While all other lesion groups performed well irrespective of delay, agranular insular cortex lesioned animals displayed mild difficulties under the no delay condition and a complete deficit when delays of 10 and 20 s were imposed.
3.3. Transfer tests Since all lesion groups were able to distinguish positive from negative trials during the original task’s no delay condition, it was expected that all groups would similarly be able to distinguish positive and negative transfer trials delivered without delay. The effects of using different cereals (transfer test) on performance on the first set of trials (Day 1) within the memory for magnitude of reinforcement task at no delay are shown for each group in Fig. 6. The graph indicates that all groups transferred to the new cereals. A two-way ANOVA with lesion groups as the between factor and positive versus negative trial type as the within factor revealed that there was a significant trial type effect F1,15 =42.9, PB 0.0001, but there was no significant difference among the four lesion groups. These results suggest that rats base their go/no-go decision on the study phase cereal’s sweetness rather than its odor, texture and/or visual characteristics.
3.4. Preference tests The effects of agranular insular cortex, pre- and infralimbic cortex, anterior cingulate cortex or control lesions on total fluid consumption for the sweetness preference test are shown in Fig. 7. The graph indicates that all groups displayed a sugar solution preference. In this experiment all animals preferred the 25% sugar solution over the 50% sugar solution. A two-way ANOVA with lesion groups as the between factor and
Fig. 6. Mean latency performance of the four lesion groups for positive and negative trials based on embedding novel cereals into the magnitude of reinforcement task.
Fig. 7. Total fluid consumption of the four lesion groups for the sweetness preference task (25 vs. 50% sucrose solutions).
sweetness preference as the within factor revealed that there was a very strong preference effect F1,15 =130.9, PB 0.0001, but no significant differences among the four groups.
4. Discussion The results indicate that rats can readily learn a short-term or working memory task based on differential sugar concentrations (reward value) of specific cereals. Even though cues such as odor, texture and brightness can also contribute to this memory, transfer tests with different cereals with comparable sugar concentrations suggest that sweetness plays a critical role in the working memory representation required to perform well in this task. Rats with lesions to the agranular insular cortex display a partial deficit in remembering magnitude of sugar reinforcement when the interval between study and test phases is brief. This impairment is greatly exacerbated with the introduction of 10 and 20 s delays between the test and study phases. Although the performance of the agranular insular cortex lesioned rats under the no delay condition is significantly worse than the other lesion groups, we consider this deficit to be partial since even these animals are still able to demonstrate substantially slower mean latencies for negative trials compared to positive trials. In fact, a 3–4 s mean latency difference between positive and negative trials (which is roughly what the agranular insular cortex lesioned rats demonstrate at no delay) is regarded as good performance by other studies utilizing similar go/no-go paradigms [7,18]. Further evidence that this deficit is incomplete comes from the transfer test results which showed the agranular insular cortex lesioned animals are able to transfer as well as the other lesion groups to a new set of cereals. Notably, no delay period was administered during the transfer test. The fact that these animals exhibit reasonable savings under the no
248
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
delay condition, along with our lab’s recent finding that agranular insular cortex lesioned rats display good motor inhibition on a go/no-go spatial continuous recognition task using an 8-arm radial maze (unpublished observations), support the notion that the total deficit observed during the delay trials is related to a deterioration in working memory rather than a failure in maintaining the go/no-go rule or a problem with response inhibition. This is an important point since damage to certain prefrontal areas of the rat have been found to influence rule learning [22] and perseverative responding [15]. Similarly, the deficit cannot be due to an inability to discriminate between tastes or a loss in taste preferences since agranular insular lesioned rats displayed the same sweetness preferences as controls using the different concentrations of sugar solutions test. An important question remains, however, as to why both the partial deficit at no delay and the complete deficit at delay manifest themselves as ‘errors of commission’ (i.e. a failure to respond correctly to negative trials). This phenomenon is commonly seen in a wide variety of tasks that utilize a go/no-go procedure [3,4,7,12,18,20] and is interpreted to be an adaptive response by an animal who is not certain how to respond. When an impaired animal is unable to recall whether or not displacing an object will result in reward, it will choose to displace the object-especially when the large payoff of a food reward when correct is weighed against the small amount of energy squandered for an error. Another reason why indecisive animals tend to displace rather than to inhibit responding could be due to a task-related bias for the former. Rats are trained to always displace objects during the study phase and to displace objects 50% of the time in the test phase. This means that when a rat sees an object, 75% of the time the correct response is to displace it. It is quite possible that the task’s inherent slant towards displacing objects may bias a confused post-surgery animal towards the same. Thus, from a cost-benefit and/or a task demand perspective, it makes sense that impairments in the present study become apparent only during negative trials. The results of the lesion analysis bare additional notable insights. First of all, the degree of cortical tissue damaged in the two animals whose lesions were immediately dorsal to the agranular insular cortex was equivalent to that of the animals included in the agranular insular cortex lesion group-yet the former displayed no deficit. Secondly, there was no overlap among the lesions of the various lesion groups. Together these findings demonstrate that the agranular insular cortex impairment in remembering magnitude of sugar reinforcement (1) imparts some degree of specificity and (2) is not simply an artifact of lesion size.
The orbitofrontal complex, a larger frontal area which includes the agranular insular cortex, has an established role in the processing of information related to affect. Damage to this neural region in both monkeys and rats produces changes in emotion and social behavior [14,25]. Furthermore, single unit recording has uncovered different classes of neurons in the monkey orbitofrontal cortex associated with reinforcement [29]. The results of the present study suggest that the agranular insular cortex may serve a unique role within this complex in mediating short-term memory for affect information. This proposal is compatible with the Goldman-Rakic [5] notion that prefrontal cortex subserves short-term working memory. Both the agranular insular and pre- and infralimbic cortices make extensive neuronal connections with the amygdala [1,6,21,27,28]. Yet, pre- and infralimbic lesioned rats display no impairment while the deficit of the agranular insular cortex lesioned animals is similar to that of amygdala lesioned rats performing the identical task [12]. These results point to the amygdala and the agranular insular cortex as two brain regions that may form a limbic-prefrontal circuit capable of representing affect attribute information. Importantly, this circuit does not appear to critically involve pre- and infralimbic cortex. The exact nature of amygdala and agranular insular cortex processing within this circuit remains unclear. It is possible that each structure may mediate distinct affect features or may contribute differentially to the data- versus knowledge-based system of memory representation. Without question, additional experiments need to be performed in order to ascertain the particular roles of each neural area in the encoding, storing, and retrieving of affect information. Recently, it has been demonstrated that medial prefrontal cortex lesions which included anterior cingulate cortex impair working memory for spatial response (egocentric) information; whereas, lesions to the preand infralimbic cortex impair working memory for spatial location and visual object information (unpublished observations). The present findings extend the notion that working memory for different attributes of memory is distributed across different neural regions within the prefrontal cortex by including the agranular insular cortex and its role in mediating the affect attribute. This role appears to be independent of anterior cingulate and pre- and infralimbic cortex, since lesions to these areas in the present study did not impair rats ability to remember sugar reinforcement magnitude. Further experiments that exclude the agranular insular cortex in mediating the spatial response and visual object attributes need to be performed, however, before a clear double dissociation can be made.
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
Acknowledgements This work was supported by NIH grant 2R01NS2077.12. The authors wish to acknowledge Robert Stoll for his invaluable assistance with data collection and analysis.
References [1] V.A. Allen, C.B. Saper, K.M. Hurley, D.F. Cechetto, Organization of visceral and limbic connections in the insular cortex of the rat, J. Comp. Neurol 311 (1991) 1–16. [2] J.J. Braun, P.S. Lasiter, S.W. Kiefer, The gustatory neocortex of the rat, Physiol Psychol. 10 (1982) 13–45. [3] S.O. Cole, Diazepam-induced impairment of a go/no go successive discrimination, Behav. Neur. Biol. 53 (1990) 371–377. [4] R.L. Doty, J.M. Risser, Influence of the D-2 dopamine receptor agonist quinpirole on the odor detection performance of rats before and after spiperone administration, Psychopharmacology 98 (1989) 310 – 315. [5] P.S. Goldman-Rakic, Circuitry of primate prefrontal cortex and regulation of behavior by representational memory, Hand. Physiol.: Nerv. Syst. 5 (1987) 373–417. [6] H.J. Groenewegen, H.W. Berendse, J.G. Wolters, A.H.M. Lohman, The anatomical relationship of the prefrontal cortex with the striapallidal system, the thalamus and the amygdala: Evidence for a parallel organization, Prog. Brain Res. 85 (1990) 95 – 117. [7] P. Jackson-Smith, R.P. Kesner, A.A. Chiba, Continuous recognition of spatial and nonspatial stimuli in hippocampal-lesioned rats, Behav. Neural Biol. 59 (1993) 107–119. [8] T. Janis, R.P. Kesner, A neural system analysis of spatial recognition memory: role of the prefrontal cortex, Soc. Neurosci. Abstr. 21 (1996) 1449. [9] R.P. Kesner, B.L. Bolland, M. Dakis, Memory for spatial locations, motor responses, and objects: triple dissociation among the hippocampus, caudate nucleus, and extrastriate visual cortex, Exp. Brain Res. 93 (1993) 462–470. [10] R.P. Kesner, B.V. Dimattia, Neurobiology of an attribute model of memory, in: A.R. Morrison, A.N. Epstein (Eds.), Progress in Psychobiology and Physiological Psychology, Academic Press, New York, 1987, pp. 207–277. [11] R.P. Kesner, M.E. Hunt, J.M. Williams, J.L. Long, Prefrontal cortex and working memory for spatial response, spatial location and visual object information in the rat, Cereb. Cortex 6 (1995) 311 – 318. [12] R.P. Kesner, J.M. Williams, Memory of magnitude of reinforcement: dissociation between the amygdala and hippocampus, Neurobiol. Learn. Mem. 64 (1995) 237–244.
.
249
[13] S.W. Kiefer, L.R. Leach, J.J. Braun, Taste agnosia following gustatory neocortex ablation: Dissociation from odor and generality across taste qualities, Behav. Neurosci. 98 (1984) 590–608. [14] B. Kolb, Social behavior of rats with chronic prefrontal lesions, J. Comp. Physiol. Psychol. 87 (1974) 466 – 474. [15] B. Kolb, A.J. Nonneman, R.K. Singh, Double dissociation of spatial impairments and perseveration following selective prefrontal lesions in rats, J. Comp. Physiol. Psychol. 87 (1974) 772 – 780. [16] P.S. Lasiter, D.A. Deems, J. Garcia, Involvement of the anterior insular neocortex in taste-potentiated odor aversion learning, Physiol. Behav. 34 (1985) 71 – 77. [17] P.S. Lasiter, D.A. Deems, R.L. Oetting, J. Garcia, Taste discriminations in rats lacking anterior insular gustatory neocortex, Physiol. Behav. 35 (1985) 277 – 285. [18] J.M. Long, R.P. Kesner, The effects of hippocampal formation and parietal cortex lesions on memory for allocentric distance in rats, Behav. Neurosci. 110 (1996) 922 – 932. [19] D.G. Mumby, J.J. Pinel, Rhinal cortex lesions and object recognition in rats, Behav. Neurosci. 108 (1994) 11 – 18. [20] R. Numan, M.P. Feloney, K.H. Pham, L.M. Tieber, Effects of medial septal lesions on an operant go/no-go delayed response alternation task in rats, Physiol. Behav. 58 (1995) 1263–1271. [21] O.P. Ottersen, Connections of the amygdala of the rat. iv: corticoamygdaloid and intraamygdaloid connections as studied with axonal transport of horseradish peroxidase, J. Comp. Neurol. 205 (1982) 30 – 48. [22] T. Otto, H. Eichenbaum, Complementary roles of the orbital prefrontal cortex and the perirhinal-entorhinal cortices in an odor-guided delayed-nonmatching-to-sample task, Behav. Neurosci. 106 (1992) 762 – 775. [23] G. Paxinos, C. Watson, The Rat Brain in Stereotaxic Coordinates, 2nd edn., Academic, New York, 1986. [24] A. Ravindranathan, P. Jackson-Smith, R.P. Kesner, Effects of perirhinal cortex and medial extrastriate visual cortex lesions on memory associated with an object continuous recognition task, Soc. Neurosci. Abstr. 18 (1992) 1058. [25] E.T. Rolls, A theory of emotion, and its application to understanding the neural basis of emotion, in: Y. Oomura (Ed.), Emotions: Neural and Chemical Theory, Japan Scientific Societies Press, Tokyo, 1986, pp. 325 – 344. [26] A. Routtenberg, M. Sloan, Self-stimulation in the frontal cortex of rattus norvegicus, Behav. Biol. 7 (1972) 567 – 572. [27] M. Sarter, H.J. Markowitsch, Convergence of basolateral amygdaloid and mediodorsal thalamic projections in different areas of the frontal cortex in the rat, Brain Res. Bull. 10 (1983) 607–622. [28] M. Takagishi, T. Chiba, Efferent projection of the infralimbic (area 25) region of the medial prefrontal cortex in the rat: An anterograde tracer PHA-L study, Brain Res. 10 (1991) 26–39. [29] S.J. Thorpe, E.T. Rolls, S. Maddison, The orbitofrontal cortex: Neuronal activity in the behaving monkey, Exp. Brain Res. 49 (1983) 93 – 115.
Research report
Short-term memory for food reward magnitude: The role of the prefrontal cortex William E. DeCoteau a, Raymond P. Kesner a,*, Joseph M. Williams b a
Department of Psychology, Uni6ersity of Utah, Salt Lake city, UT 84112, USA b Department of Psychology, Ohio State Uni6ersity, Columbus, OH, USA
Received 30 October 1996; received in revised form 26 February 1997; accepted 27 February 1997
Abstract Memory for magnitude of reinforcement was assessed in rats using a go/no-go short-term memory paradigm. During the task’s study phase rats were given a piece of cereal comprised of either 25 or 50% sugar. For all trials, one of the cereal types was designated positive, the other negative. On the ensuing test phase the rat was presented with an object which covered a food well. If a positive food reward was given during the study phase, a second food reward was placed beneath the object. No food reward was placed under the object if the study phase consisted of a negative food reward. Latency to object displacement was used as the measure of performance. Following the establishment of a significant difference between latency to approach the object with reward compared to latency to approach the object without reward, rats were given either agranular insular cortex, anterior cingulate cortex, pre- and infralimbic cortex or control lesions. Agranular insular cortex lesioned animals demonstrated a mild post-surgery impairment. Trials consisting of 10 and 20 s delays between the study and test phases were then introduced. Delays exacerbated the previous deficit of the agranular insular cortex lesion group, but had little effect on the other lesion groups. All animals transferred to a new set of cereals containing 25 and 50% sugar and exhibited taste preferences to sugar solutions of different concentrations. These results indicate that the agranular insular cortex may play an important role in the processing of affect-laden information within a prefrontal cortex short-term or working memory system. © 1997 Elsevier Science B.V. Keywords: Prefrontal cortex; Agranular insular cortex; Working memory; Reward; Rat
1. Introduction Kesner has proposed that new information is represented within a data-based memory system by multiple forms or attributes of memory [10]. In support of this proposal it has been demonstrated that in tasks that require novel and/or unique information for each trial, the hippocampus mediates short-term memory for spatial and temporal attributes, the extrastriate and perirhinal cortex mediate short-term memory for visual object information as an example of a sensory-per* Corresponding author. Tel.: + 1 801 581 7430; fax: +1 801 581 5841; e-mail: [email protected] 0166-4328/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S 0 1 6 6 - 4 3 2 8 ( 9 7 ) 0 0 0 4 4 - 2
ceptual attribute, the caudate-putamen mediates shortterm memory for a response attribute and amygdala mediates short-term memory for an affect attribute [9,12,19,24]. In recent years it has become clear that there is not just one neural region, but rather a neural circuit, that mediates short-term memory for specific attributes within the data-based memory system. For example, for the spatial attribute short-term memory representations can be found in the pre- and parasubiculum, medial entorhinal cortex and pre- and infralimbic cortex in addition to the hippocampus [8] (and unpublished observations). Similarly, it is highly likely that a neural circuit mediates the affect attribute as measured by memory for magnitude of reinforce-
240
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
ment. To test this idea the present study is focused on the agranular insular cortex. This structure was chosen because it has been shown to support intracranial selfstimulation [26], lesions to the area, when combined with adjacent gustatory cortex, produce impairments in taste aversion learning [2,13,17] and taste-potentiated odor aversion [16], and because of its direct neural connections with the amygdala [1,6,21,27]. Thus, the first purpose of this study is to examine the role of the agranular insular cortex in a successive delayed conditional discrimination task for magnitude of reward value. In the study phase of the task, rats were given one of two cereals. One cereal contained 25% sugar; the other 50% sugar. One of the two cereals was always designated as the positive stimulus and the other as the negative stimulus. This study phase was followed by the test phase in which the rat was shown an object which covered a food well. If the rat was given the positive food stimulus during the study phase, another food reward was placed beneath the object. Rats learn to approach the objects quickly when they expect a reward and they are slow to approach the object when they expect no reward. Recent research using this task has demonstrated that rats with amygdala lesions are impaired in short-term memory for reward value, even though they display normal preference for differential reward [12]. In previous work, it has been shown that short-term or working memory for different forms or attributes of memory is distributed across different neural regions within the prefrontal cortex of the rat. For example, pre- and infralimbic cortex mediates short-term memory for visual object and spatial attributes, whereas the anterior cingulate cortex and medial precentral cortex mediate the response attribute [11]. In order to determine whether short-term memory for the affect attribute is uniquely subserved by the agranular insular cortex, the second purpose of this study is to examine pre- and infralimbic cortex and the anterior cingulate cortex to ascertain the role of these prefrontal structures in mediating the affect attribute. The pre- and infralimbic cortical area is of particular interest since, like the agranular insular cortex, it has direct reciprocal connections with the amygdala [27,28].
2. Materials and methods
2.1. Subjects Nineteen male Hooded Long-Evans rats, weighing 300 – 350 g at the start of the experiment, were used as subjects. They were housed in standard stainless steel cages in a large, well lit laboratory room and were maintained on a 12:12 h light:dark schedule. All animals were placed on food deprivation with ad lib water
and maintained at 80–85% of ad lib weight throughout the experiment.
2.2. Apparatus The apparatus consisted of a platform, 122.0 cm long by 35.5 cm wide, with a black Plexiglas door (50.5 cm tall by 43.0 cm wide) separating the table into two halves. A pulley system enabled the experimenter to raise or lower the black Plexiglas door. The table was constructed of painted black wood and was raised 91.0 cm above the floor. Each side contained three 2.6 cm diameter food wells, separated by 6.0 cm, located 5.0 cm from the back edge, and centered from side-to-side. A sheet of red Plexiglas (88.0 cm tall by 91.0 cm wide) extended across one side of the table to block the animal’s view of the experimenter and the room; the opposite side of the table was against a wall. The stimuli for this experiment were a variety of three dimensional objects.
2.3. Procedure Rats were given a 3 week pre-training period. Rats initially learned to obtain food from the food wells. They were subsequently conditioned to push aside a 3-dimensional object covering the food well in order to obtain a food reward. During this pre-training phase, rats were habituated to three cereals: Oat Bake which contains 25% sugar; Froot Loops which contains 50% sugar and Apple Jacks (50% sugar) which was to be used as a neutral food stimulus. Once the rats learned to uncover the food wells, the training sessions began. The rats always began each trial at the same end of the apparatus. At the start of a trial (the study phase) the black Plexiglas door was raised allowing the rat to enter the other side of the apparatus. The door was lowered after the rat crossed to the other side. In the study phase of the trial, the rat encountered a 3-dimensional object covering the center food well. The objects used for each rat remained constant throughout the experiment. Located in the food well was one of two cereals: either an Oat Bake or a Froot Loop. One of the two cereals was always designated as the positive stimulus and the other was always designated as the negative stimulus. This assignment was chosen randomly but remained constant throughout the testing period for each rat. Each rat received ten trials per day, 5 days per week. Five trials were designated as positive and five trials were designated as negative. After the rat consumed the cereal, the opaque Plexiglas door was again raised allowing the rat access to the other side of the apparatus. This constituted the start of the test phase. During the test phase, the rat encountered a different object covering the center food well. If
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
the rat received the positive stimulus cereal during the study phase, in the subsequent test phase, another food reward (the neutral cereal) was placed beneath the object. If the rat received the negative stimulus during the study phase, no food reward was placed in the center food well during the subsequent test phase. Latency measured from crossing the center to the displacement of the object to uncover the center food well during the test phase was used as the dependent measure. The trial ended when the rat displaced the object from over the foodwell or ten seconds had elapsed.
2.4. Surgery After rats reached a criterion level of at least a 4.5 s mean latency difference between the positive and negative trials for two consecutive blocks of 50 trials, they were given either agranular insular cortex, pre- and infralimbic cortex, anterior cingulate cortex, or sham operated lesions. There were seven rats in the agranular insular cortex group, and four rats each in the pre- and infralimbic cortex, anterior cingulate cortex, and sham control groups. All surgeries were performed under sodium pentobarbital anesthesia (50 mg/kg, ip). All lesion sites were based on the Paxinos and Watson [23] rat stereotaxic atlas. Lesions to the agranular insular cortex, and pre- and infralimbic cortex were made electrolytically using stainless steel electrodes that were insulated except for a 0.5 mm tip. Direct current (1.2 mA) was applied for 15 s for each agranular insular cortex electrode placement and for 10 s for each preand infralimbic cortex placement. The coordinates for the agranular insular cortex lesion were as follows: 4.5 mm posterior to bregma, 3.2 mm lateral to midline, 2.4 mm ventral to dura; 3.5 mm posterior to bregma, 4.3 mm lateral to midline, 3.8 mm ventral to dura; and 2.5 mm posterior to bregma, 5.0 mm lateral to midline, 3.8 mm ventral to dura. Pre- and infralimbic lesions were made using the following coordinates: 3.2 mm posterior to bregma, 0.5 mm lateral to midline, 2.5 and 4.2 mm ventral to dura; and 2.2 mm posterior to bregma, 0.5 mm lateral to midline, 3.1 mm ventral to dura. Anterior cingulate cortex lesioned rats had the bone and dura removed above the lesion site. The region 0.5 mm anterior to bregma to the frontal pole, 1.0 lateral to midline, and 3.0 mm ventral to dura was then aspirated. All rats were then provided with a 1 week recuperative period. Post-surgery testing involved 200 trials on the same task learned prior to surgery. Following these 200 post-surgery trials, rats were tested at delays of 10 and 20 s between the study phase and the test phase. Rats received 100 trials each at 10 s and at 20 s delays. Subsequent to the 20 s delay testing, rats were returned to the no delay condition and given transfer tests to assess whether rats were using odor, texture
241
and/or visual cues to solve the task rather than sweetness differences between the two cereals. Rats received the same ten trials per day as before. However, for two of the ten trials, a novel set of cereals was substituted. The new cereals contained approximately the same percentage of sugar as the original cereals. Almond Delight, which contains approximately 25% sugar, was used as a substitute for Oat Bake. Kellogg’s Honey Smacks, which contain approximately 50% sugar, was used as a substitute for Froot Loops. As a final control, rats were given a preference task to determine whether impairments on the original task were the result of difficulty with memory for sweetness intensity rather than an inability to discriminate between sweetness intensity. In the preference task, the rats were allowed 15 min of water per day and an unlimited supply of food. Prior to testing, rats were habituated to 15 min of water per day. Following habituation, rats alternated between days where they were presented with two bottles, one containing a 25%, the other a 50% sugar solution, and days where they were given two bottles containing only water. The water condition served as a control for the sugar solution condition. To prevent any side bias the placement of each sugar solution bottle was randomly located to the left or right during each day of testing. The observer recorded the amount of water the rat drank from each bottle. Rats were tested on this task for eight consecutive days.
2.5. Histological 6erification Following these tests, rats were sacrificed with an overdose of sodium pentobarbital (50 mg/kg, ip) and perfused with 10% formalin. The brains were removed and stored in 30% sucrose formalin for 1 week. They were then frozen for slide preparation. Transverse sections (24 mm) were cut with a cryostat through the lesion area and stained with cresyl violet. Brain sections were examined for histological accuracy of the lesion placement.
3. Results
3.1. Histology Histological analysis of the agranular insular cortex lesioned brains revealed that two animals had damage immediately dorsal to the intended site and were thus excluded from the agranular insular cortex group and added to the sham control group. For these two animals the anterior portion of their lesions consisted of bilateral destruction to the lateral frontal cortex, area 2, and parietal cortex, area 1, and the posterior aspect of their lesions damaged granular and dysgranular insular
242
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
Fig. 1. Illustrations of the smallest (black) and largest (stippled) agranular insular cortex lesions.
cortex. The overall degree of cortical tissue damage in these animals was very similar to the amount lost in the animals included in the agranular insular cortex lesioned group. Representative examples of the smallest
and largest of the five remaining agranular insular cortex lesions are shown in Fig. 1. Examination of these animals indicated that they all received bilateral removal along the entire anterior-posterior axis of the
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
243
Fig. 2. Illustrations of the smallest (black) and largest (stippled) pre- and infralimbic cortex lesions.
agranular insular cortex. The anterior components of the larger lesions contained some damage to the ventral lateral portions of frontal cortex, areas 2 and 3. The posterior aspect of the larger lesions included damage to ventral parietal cortex, area 1, and to the dysgranular insular cortex. Granular insular cortex was consistently spared in all agranular insular cortex lesions.
Figs. 2 and 3 show representative examples of the smallest and largest pre- and infralimbic and anterior cingulate lesions, respectively. Histological analysis of the pre- and infralimbic cortex lesioned brains revealed bilateral destruction of the pre- and infralimbic cortex. Larger lesions included additional damage to the posterior aspect of ventromedial orbital cortex, cingulate cortex, area 3 and the medial aspect of the caudate-
244
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
Fig. 3. Illustrations of the smallest (black) and largest (stippled) anterior cingulate cortex lesions.
putamen. All anterior cingulate cortex lesions included damage to frontal cortex, area 2 and cingulate cortex, area 1. The anterior portion of the lesion included removal of cingulate cortex, area 3.
3.2. Memory for magnitude of reinforcement Analyses of the two animals excluded on histological grounds from the agranular insular cortex group indi-
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
245
Fig. 4. Pre-and post-operative mean latency for positive and negative trials for control. (a), agranular insular cortex (b), pre- and infralimbic cortex (c) and anterior cingulate cortex (d) lesioned animals.
cated that they were not significantly different from the sham control group in either pre- or post-surgery performance. The animals were thus combined to form one control group (n=6) for the remainder of the analyses. The mean number of trials to learn this task was 475 for the control group, 480 for the agranular insular cortex group, 513 for the pre- and infralimbic cortex group and 538 for the anterior cingulate cortex group. A one-way analysis of variance (ANOVA) indicated that the differences among the four groups were not significant. The effects of the four types of lesions on pre- and post-operative mean latency for positive and negative trials are shown in Fig. 4 (a,b,c and d). The data were grouped into blocks of 100 trials for analysis. This included the criterion trials (Pre) and 2 sets of post surgery trials (Post 1 and Post 2). A repeated measures
three-way ANOVA with lesion groups as the between factor and pre-post 1-post 2 (blocks) and positive-negative trial type as the within factors revealed that there was a significant lesion effect, F3,15 = 10.3, PB0.001, a significant block effect, F2,30 = 3.7, PB 0.05, a significant trial type effect, F1,15 = 545.0, PB0.0001, a significant block by lesion interaction effect, F6,30 =3.1, PB 0.01, a significant trial type by lesion interaction effect, F3,15 = 4.4, PB 0.05, and a significant block by trial type interaction effect, F2,30 = 4.6, PB .01. Further Newman-Keuls comparison tests of the block by lesion interaction indicated that there were no significant differences among the four groups with respect to pre-operative performance and the pre- and infralimbic cortex, anterior cingulate cortex and cortical control animals showed no significant deficit in their post-operative compared to pre-operative performance. The
246
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
Fig. 5. Mean latencies during 1–4, 10, and 20 s delay conditions for positive and negative trials for control (a), agranular insular cortex (b), preand infralimbic cortex (c) and anterior cingulate cortex (d) lesioned animals.
agranular insular cortex lesioned group, however, had a deficit in performance in both the first and second post-operative blocks compared to pre-operative performance (P B 0.05). Newman-Keuls comparison tests of the trial type by lesion interaction revealed that the impairment of the agranular insular cortex group was due to poor performance on negative trials (P B0.05). The effects of agranular insular cortex, pre- and infralimbic cortex, anterior cingulate cortex or control lesions on mean latency performance for the positive and negative trials at 1 – 4, 10, and 20 s delay conditions are shown in Fig. 5 (a – d). Data were again grouped into blocks of 100 trials for analysis which included the last 100 trials of the immediate (1 – 4 s) condition and 100 trials each for the 10 and 20 sec conditions. A repeated measures three-way ANOVA with lesion groups as the between factor and delays and trial type as the within factors revealed that there was a signifi-
cant lesion effect F3,15 = 20.2, PB 0.0001, a significant block effect F2,30 = 7.6, PB 0.005, a significant trial type effect F1,15 = 237.6, PB0.0001, a significant block by trial type interaction effect F2,30 = 6.4, PB0.005, and a significant trial type by lesion interaction effect F3,15 = 8.9, P B0.001. Further Newman-Keuls comparison tests of the trial type by lesion interaction indicated that the agranular insular cortex group was impaired compared to the other three lesion groups and that this deficit was due to poor performance on negative trials (PB 0.05). It is important to note that, in observing the behavior of the animals executing the task, it was clear that the poor performance of the agranular insular lesioned animals during both the no delay and delay conditions was not due to a change in running speed. Immediately upon the lifting of the guillotine door the animals of all lesion groups would quickly run the length of appara-
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
247
tus. It was not until they were within its close proximity that the rats decided whether or not to displace the test phase object. At this moment impairments became observable. While all other lesion groups performed well irrespective of delay, agranular insular cortex lesioned animals displayed mild difficulties under the no delay condition and a complete deficit when delays of 10 and 20 s were imposed.
3.3. Transfer tests Since all lesion groups were able to distinguish positive from negative trials during the original task’s no delay condition, it was expected that all groups would similarly be able to distinguish positive and negative transfer trials delivered without delay. The effects of using different cereals (transfer test) on performance on the first set of trials (Day 1) within the memory for magnitude of reinforcement task at no delay are shown for each group in Fig. 6. The graph indicates that all groups transferred to the new cereals. A two-way ANOVA with lesion groups as the between factor and positive versus negative trial type as the within factor revealed that there was a significant trial type effect F1,15 =42.9, PB 0.0001, but there was no significant difference among the four lesion groups. These results suggest that rats base their go/no-go decision on the study phase cereal’s sweetness rather than its odor, texture and/or visual characteristics.
3.4. Preference tests The effects of agranular insular cortex, pre- and infralimbic cortex, anterior cingulate cortex or control lesions on total fluid consumption for the sweetness preference test are shown in Fig. 7. The graph indicates that all groups displayed a sugar solution preference. In this experiment all animals preferred the 25% sugar solution over the 50% sugar solution. A two-way ANOVA with lesion groups as the between factor and
Fig. 6. Mean latency performance of the four lesion groups for positive and negative trials based on embedding novel cereals into the magnitude of reinforcement task.
Fig. 7. Total fluid consumption of the four lesion groups for the sweetness preference task (25 vs. 50% sucrose solutions).
sweetness preference as the within factor revealed that there was a very strong preference effect F1,15 =130.9, PB 0.0001, but no significant differences among the four groups.
4. Discussion The results indicate that rats can readily learn a short-term or working memory task based on differential sugar concentrations (reward value) of specific cereals. Even though cues such as odor, texture and brightness can also contribute to this memory, transfer tests with different cereals with comparable sugar concentrations suggest that sweetness plays a critical role in the working memory representation required to perform well in this task. Rats with lesions to the agranular insular cortex display a partial deficit in remembering magnitude of sugar reinforcement when the interval between study and test phases is brief. This impairment is greatly exacerbated with the introduction of 10 and 20 s delays between the test and study phases. Although the performance of the agranular insular cortex lesioned rats under the no delay condition is significantly worse than the other lesion groups, we consider this deficit to be partial since even these animals are still able to demonstrate substantially slower mean latencies for negative trials compared to positive trials. In fact, a 3–4 s mean latency difference between positive and negative trials (which is roughly what the agranular insular cortex lesioned rats demonstrate at no delay) is regarded as good performance by other studies utilizing similar go/no-go paradigms [7,18]. Further evidence that this deficit is incomplete comes from the transfer test results which showed the agranular insular cortex lesioned animals are able to transfer as well as the other lesion groups to a new set of cereals. Notably, no delay period was administered during the transfer test. The fact that these animals exhibit reasonable savings under the no
248
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
delay condition, along with our lab’s recent finding that agranular insular cortex lesioned rats display good motor inhibition on a go/no-go spatial continuous recognition task using an 8-arm radial maze (unpublished observations), support the notion that the total deficit observed during the delay trials is related to a deterioration in working memory rather than a failure in maintaining the go/no-go rule or a problem with response inhibition. This is an important point since damage to certain prefrontal areas of the rat have been found to influence rule learning [22] and perseverative responding [15]. Similarly, the deficit cannot be due to an inability to discriminate between tastes or a loss in taste preferences since agranular insular lesioned rats displayed the same sweetness preferences as controls using the different concentrations of sugar solutions test. An important question remains, however, as to why both the partial deficit at no delay and the complete deficit at delay manifest themselves as ‘errors of commission’ (i.e. a failure to respond correctly to negative trials). This phenomenon is commonly seen in a wide variety of tasks that utilize a go/no-go procedure [3,4,7,12,18,20] and is interpreted to be an adaptive response by an animal who is not certain how to respond. When an impaired animal is unable to recall whether or not displacing an object will result in reward, it will choose to displace the object-especially when the large payoff of a food reward when correct is weighed against the small amount of energy squandered for an error. Another reason why indecisive animals tend to displace rather than to inhibit responding could be due to a task-related bias for the former. Rats are trained to always displace objects during the study phase and to displace objects 50% of the time in the test phase. This means that when a rat sees an object, 75% of the time the correct response is to displace it. It is quite possible that the task’s inherent slant towards displacing objects may bias a confused post-surgery animal towards the same. Thus, from a cost-benefit and/or a task demand perspective, it makes sense that impairments in the present study become apparent only during negative trials. The results of the lesion analysis bare additional notable insights. First of all, the degree of cortical tissue damaged in the two animals whose lesions were immediately dorsal to the agranular insular cortex was equivalent to that of the animals included in the agranular insular cortex lesion group-yet the former displayed no deficit. Secondly, there was no overlap among the lesions of the various lesion groups. Together these findings demonstrate that the agranular insular cortex impairment in remembering magnitude of sugar reinforcement (1) imparts some degree of specificity and (2) is not simply an artifact of lesion size.
The orbitofrontal complex, a larger frontal area which includes the agranular insular cortex, has an established role in the processing of information related to affect. Damage to this neural region in both monkeys and rats produces changes in emotion and social behavior [14,25]. Furthermore, single unit recording has uncovered different classes of neurons in the monkey orbitofrontal cortex associated with reinforcement [29]. The results of the present study suggest that the agranular insular cortex may serve a unique role within this complex in mediating short-term memory for affect information. This proposal is compatible with the Goldman-Rakic [5] notion that prefrontal cortex subserves short-term working memory. Both the agranular insular and pre- and infralimbic cortices make extensive neuronal connections with the amygdala [1,6,21,27,28]. Yet, pre- and infralimbic lesioned rats display no impairment while the deficit of the agranular insular cortex lesioned animals is similar to that of amygdala lesioned rats performing the identical task [12]. These results point to the amygdala and the agranular insular cortex as two brain regions that may form a limbic-prefrontal circuit capable of representing affect attribute information. Importantly, this circuit does not appear to critically involve pre- and infralimbic cortex. The exact nature of amygdala and agranular insular cortex processing within this circuit remains unclear. It is possible that each structure may mediate distinct affect features or may contribute differentially to the data- versus knowledge-based system of memory representation. Without question, additional experiments need to be performed in order to ascertain the particular roles of each neural area in the encoding, storing, and retrieving of affect information. Recently, it has been demonstrated that medial prefrontal cortex lesions which included anterior cingulate cortex impair working memory for spatial response (egocentric) information; whereas, lesions to the preand infralimbic cortex impair working memory for spatial location and visual object information (unpublished observations). The present findings extend the notion that working memory for different attributes of memory is distributed across different neural regions within the prefrontal cortex by including the agranular insular cortex and its role in mediating the affect attribute. This role appears to be independent of anterior cingulate and pre- and infralimbic cortex, since lesions to these areas in the present study did not impair rats ability to remember sugar reinforcement magnitude. Further experiments that exclude the agranular insular cortex in mediating the spatial response and visual object attributes need to be performed, however, before a clear double dissociation can be made.
W.E. DeCoteau et al. / Beha6ioural Brain Research 88 (1997) 239–249
Acknowledgements This work was supported by NIH grant 2R01NS2077.12. The authors wish to acknowledge Robert Stoll for his invaluable assistance with data collection and analysis.
References [1] V.A. Allen, C.B. Saper, K.M. Hurley, D.F. Cechetto, Organization of visceral and limbic connections in the insular cortex of the rat, J. Comp. Neurol 311 (1991) 1–16. [2] J.J. Braun, P.S. Lasiter, S.W. Kiefer, The gustatory neocortex of the rat, Physiol Psychol. 10 (1982) 13–45. [3] S.O. Cole, Diazepam-induced impairment of a go/no go successive discrimination, Behav. Neur. Biol. 53 (1990) 371–377. [4] R.L. Doty, J.M. Risser, Influence of the D-2 dopamine receptor agonist quinpirole on the odor detection performance of rats before and after spiperone administration, Psychopharmacology 98 (1989) 310 – 315. [5] P.S. Goldman-Rakic, Circuitry of primate prefrontal cortex and regulation of behavior by representational memory, Hand. Physiol.: Nerv. Syst. 5 (1987) 373–417. [6] H.J. Groenewegen, H.W. Berendse, J.G. Wolters, A.H.M. Lohman, The anatomical relationship of the prefrontal cortex with the striapallidal system, the thalamus and the amygdala: Evidence for a parallel organization, Prog. Brain Res. 85 (1990) 95 – 117. [7] P. Jackson-Smith, R.P. Kesner, A.A. Chiba, Continuous recognition of spatial and nonspatial stimuli in hippocampal-lesioned rats, Behav. Neural Biol. 59 (1993) 107–119. [8] T. Janis, R.P. Kesner, A neural system analysis of spatial recognition memory: role of the prefrontal cortex, Soc. Neurosci. Abstr. 21 (1996) 1449. [9] R.P. Kesner, B.L. Bolland, M. Dakis, Memory for spatial locations, motor responses, and objects: triple dissociation among the hippocampus, caudate nucleus, and extrastriate visual cortex, Exp. Brain Res. 93 (1993) 462–470. [10] R.P. Kesner, B.V. Dimattia, Neurobiology of an attribute model of memory, in: A.R. Morrison, A.N. Epstein (Eds.), Progress in Psychobiology and Physiological Psychology, Academic Press, New York, 1987, pp. 207–277. [11] R.P. Kesner, M.E. Hunt, J.M. Williams, J.L. Long, Prefrontal cortex and working memory for spatial response, spatial location and visual object information in the rat, Cereb. Cortex 6 (1995) 311 – 318. [12] R.P. Kesner, J.M. Williams, Memory of magnitude of reinforcement: dissociation between the amygdala and hippocampus, Neurobiol. Learn. Mem. 64 (1995) 237–244.
.
249
[13] S.W. Kiefer, L.R. Leach, J.J. Braun, Taste agnosia following gustatory neocortex ablation: Dissociation from odor and generality across taste qualities, Behav. Neurosci. 98 (1984) 590–608. [14] B. Kolb, Social behavior of rats with chronic prefrontal lesions, J. Comp. Physiol. Psychol. 87 (1974) 466 – 474. [15] B. Kolb, A.J. Nonneman, R.K. Singh, Double dissociation of spatial impairments and perseveration following selective prefrontal lesions in rats, J. Comp. Physiol. Psychol. 87 (1974) 772 – 780. [16] P.S. Lasiter, D.A. Deems, J. Garcia, Involvement of the anterior insular neocortex in taste-potentiated odor aversion learning, Physiol. Behav. 34 (1985) 71 – 77. [17] P.S. Lasiter, D.A. Deems, R.L. Oetting, J. Garcia, Taste discriminations in rats lacking anterior insular gustatory neocortex, Physiol. Behav. 35 (1985) 277 – 285. [18] J.M. Long, R.P. Kesner, The effects of hippocampal formation and parietal cortex lesions on memory for allocentric distance in rats, Behav. Neurosci. 110 (1996) 922 – 932. [19] D.G. Mumby, J.J. Pinel, Rhinal cortex lesions and object recognition in rats, Behav. Neurosci. 108 (1994) 11 – 18. [20] R. Numan, M.P. Feloney, K.H. Pham, L.M. Tieber, Effects of medial septal lesions on an operant go/no-go delayed response alternation task in rats, Physiol. Behav. 58 (1995) 1263–1271. [21] O.P. Ottersen, Connections of the amygdala of the rat. iv: corticoamygdaloid and intraamygdaloid connections as studied with axonal transport of horseradish peroxidase, J. Comp. Neurol. 205 (1982) 30 – 48. [22] T. Otto, H. Eichenbaum, Complementary roles of the orbital prefrontal cortex and the perirhinal-entorhinal cortices in an odor-guided delayed-nonmatching-to-sample task, Behav. Neurosci. 106 (1992) 762 – 775. [23] G. Paxinos, C. Watson, The Rat Brain in Stereotaxic Coordinates, 2nd edn., Academic, New York, 1986. [24] A. Ravindranathan, P. Jackson-Smith, R.P. Kesner, Effects of perirhinal cortex and medial extrastriate visual cortex lesions on memory associated with an object continuous recognition task, Soc. Neurosci. Abstr. 18 (1992) 1058. [25] E.T. Rolls, A theory of emotion, and its application to understanding the neural basis of emotion, in: Y. Oomura (Ed.), Emotions: Neural and Chemical Theory, Japan Scientific Societies Press, Tokyo, 1986, pp. 325 – 344. [26] A. Routtenberg, M. Sloan, Self-stimulation in the frontal cortex of rattus norvegicus, Behav. Biol. 7 (1972) 567 – 572. [27] M. Sarter, H.J. Markowitsch, Convergence of basolateral amygdaloid and mediodorsal thalamic projections in different areas of the frontal cortex in the rat, Brain Res. Bull. 10 (1983) 607–622. [28] M. Takagishi, T. Chiba, Efferent projection of the infralimbic (area 25) region of the medial prefrontal cortex in the rat: An anterograde tracer PHA-L study, Brain Res. 10 (1991) 26–39. [29] S.J. Thorpe, E.T. Rolls, S. Maddison, The orbitofrontal cortex: Neuronal activity in the behaving monkey, Exp. Brain Res. 49 (1983) 93 – 115.