Learning and Recall of Form Discriminations During Reversible Cooling Deactivation of Ventral-Posterior Suprasylvian Cortex In

Proc. Natl. Acad. Sci. USA Vol. 93, pp. 1654-1658, February 1996 Neurobiology

Learning and recall of form discriminations during reversible cooling deactivation of ventral-posterior suprasylvian cortex in the cat (memory/object and pattern discriminations/area 20/inferotemporal cortex/monkey) STEPHEN G. LOMBER*t, BERTRAM R. PAYNE*, AND PAUL CORNWELL*t *Laboratory for Visual Perception and Cognition, Department of Anatomy and Neurobiology, Boston University School of Medicine, 700 Albany Street, Boston, MA 02118; and tDepartment of Psychology, 130 Moore Building, The Pennsylvania State University, University Park, PA 16802 Communicated by Irving T. Diamond, Duke University Medical Center, Durham, NC, October 31, 1995

ABSTRACT Extrastriate visual cortex of the ventral- posterior suprasylvian gyrus and the fusiform gyrus. The cats' posterior suprasylvian gyrus (vPS cortex) of freely behaving ability to discriminate pattern or object pairs was tested while cats was reversibly deactivated with cooling to determine its cortex was warm (active) and cool (deactivated). We examined role in performance on a battery of simple or masked two- discrimination performance on four tasks: (i) highly familiar dimensional pattern discriminations, and three-dimensional simple two-dimensional patterns; (ii) highly familiar masked or object discriminations. Deactivation of vPS cortex by cooling partially occluded patterns; (iii) highly familiar three- profoundly impaired the ability of the cats to recall the dimensional objects; and (iv) recently learned or novel objects. difference between all previously learned pattern and object Apparatus and Method. A two-choice Pennsylvania General discriminations. However, the cats' ability to learn or relearn Testing Apparatus (PGTA) was used for all discrimination pattern and object discriminations while vPS was deactivated training and testing (7, 8). Detailed procedures for apparatus depended upon the nature ofthe pattern or object and the cats' acclimation and training are described elsewhere (7). Patterns prior level of exposure to them. During cooling of vPS cortex, were displayed on video monitors in each goal compartment. the cats could neither learn the novel object discriminations For object discriminations, the video displays were removed, nor relearn a highly familiar masked or partially occluded and objects were centrally located in the distal half of each goal pattern discrimination, although they could relearn both the compartment. A discrimination trial began with the animal in highly familiar object and simple pattern discriminations. the start box and the raising of an opaque door for the cat to These cooling-induced deficits resemble those induced by view the discriminanda through a transparent door. Two cooling of the topologically equivalent inferotemporal cortex seconds later, the transparent door was raised and the cat of monkeys and provides evidence that the equivalent regions entered the decision area. The cat was deemed to have made contribute to visual processing in similar ways. a decision when both front paws were set down across a line separating the decision area from the goal compartments. For In mammals, both the learning and recall of pattern and object responding to the positive stimulus, cats were rewarded with discriminations depends upon cerebral cortex. In monkeys, the soft commercial cat food, which they found behind a lip above ability to discriminate between pairs of patterns or objects has the video monitor or behind the object. been localized and depends upon inferotemporal (IT) cortex Training. Initial training was done binocularly and the cats (1). In cats, the topologically equivalent cortex (2) comprises were taught to discriminate between a solid black "I" and "O"U the ventral half of the posterior suprasylvian gyrus (vPS object, simple patterns (outline I and 0), and masked patterns cortex), and it is known that this region contributes to the (outline I and 0 masked by multiple intersecting diagonal lines); learning of some pattern discriminations (3-5), but no previ- to criterion levels of performance (92%; 23 correct responses out ous investigation has identified any clear contribution of vPS of 25 consecutive trials). After reaching criterion, the cats con- cortex to the learning or recall of objects or masked patterns. tinued to be trained on each I vs. 0 discrimination for an These shortfalls in our knowledge of vPS cortex have made it additional 5000 trials to ensure the cats were highly familiar with difficult to determine whether cat vPS cortex has more than the discriminations. For discriminations designed to test learning a superficial functional similarity to monkey IT cortex (2). of new discriminations, the cats were initially trained to discrim- We attempted to clarify this relationship by examining the inate multiple pairs ofjunk objects (>20) prior to exposure to the abilities of cats to learn and recall a battery of pattern and tested object pairs. Recall of these object discriminations was object discriminations while the vPS cortex was deactivated examined 4-18 hr later in a single block of 25 trials. by cooling. Surgical Procedures. Following initial training, the cats underwent three surgical procedures to: (i) section the optic MATERIALS AND METHODS chiasm in the midline by a transoral approach to allow visual signals to be directed to one hemisphere when the contralat- Two laboratory-reared cats [Toonces (Cat T) and Sid (Cat Sd)] eral eye was occluded; (ii) transect the visual fibers in the were studied and treated in accordance with the National corpus callosum (9) to eliminate possible direct interactions Institutes of Health Guide for the Care and Use of Laboratory between visual cortices in the two hemispheres (10); and (iii) Animals (no. 86-23). The cats were trained on a series of install subdural cortical cooling probes over vPS cortex of the two-choice pattern- or object-discriminations prior to midline right hemisphere. All procedures were performed by antiseptic sections of the optic chiasm and the caudal two-thirds of the methods and with the cats fully anesthetized with sodium corpus callosum. Cooling probes (6) were permanently im- pentobarbital (40 mg/kg i.p. and i.v. as necessary). The planted in contact with cortex of the ventral half of the right completeness of the optic chiasm section was verified following

The publication costs of this article were defrayed in part by page charge Abbreviations: vPS cortex, ventral-posterior suprasylvian gyrus; IT, payment. This article must therefore be hereby marked "advertisement" in inferotemporal. accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed. 1654 Downloaded by guest on September 29, 2021 Neurobiology: Lomber et al. Proc. Natl. Acad. Sci. USA 93 (1996) 1655

surgery by the complete loss of the peripheral visual fields in v psss a visual perimetry test. Cooling Probes. Loops of 23-gauge hypodermic tubing were shaped to conform to the shape of vPS cortex and the fusiform 5. -A gyrus. Cooling was effected by pumping cold methanol through the hypodermic tubing. Temperature of the cooling probe was monitored continuously via a thermocouple at- tached to the probe and was kept constant (±+1C) by adjusting the methanol flow rate. Following probe implantation, the cats wore a harness during testing and were loosely tethered to an overhead bar. The tether supported the cooling tubes and temperature monitoring wire and was long enough to afford the animal complete freedom of movement within the apparatus. Testing. The cats' capacity to recall the highly familiar patterns and objects was tested while the cortex either was at normal body temperature and active or was cooled and FIG. 1. Horizontal section through vPS cortex stained for Nissl deactivated. Testing consisted of a single 25-trial block in each substance. Measurement sites (arrowheads) are identified and tem- of three to six individual sessions. In the next series, which peratures are indicated. The probe was cooled to 3°C. The black line continued for the next 4 days, the cat's capacity to relearn the through cortex estimates the position of the 20°C thermocline and same highly familiar pattern or object discriminations was separates the deactivated regions close to the pial surface from the tested until either criterion performance was reached or active regions either close to white matter or some distance away from 100-150 trials were performed on a given day. During this the cooled region. Two small nicks in layer I resulted during tissue testing, the vPS probe was cooled to 3 ± PC. On day 5, the cats' processing. A, anterior; H, hippocampal formation; L, lateral; psss, capacity to relearn the discrimination was tested with vPS posterior suprasylvian sulcus; T, tentorial surface; V, lateral ventricle. cortex at normal body temperature and active. In the final (Bar = 3 mm.) series, the cats' capacity to learn novel object discriminations conduction, and there were certainly no direct effects of was also tested while the vPS cortex was cooled. Since the cats cooling on neural processing in hippocampus and nearby showed no evidence of being able to learn an object discrim- structures. ination with vPS cortex cooled, they were then trained with Behavior. Detailed data on recall and (re)learning of various cortex warm. Subsequently, their capacity to recall the same form discriminations from Cats T and Sd are presented in Fig. 3. discrimination was tested 4-18 hr later in two blocks of 25 Recall. vPS cortex is critical for the recall of both trials. During the first block of trials vPS cortex was cooled and highly deactivated, familiar and recently learned form discriminations. In each cat, and in the second block, vPS cortex was warm. and for both pattern and object discriminations, cooling of the Final Procedures. At the conclusion of behavioral testing, vPS cortex completely the cats were anesthetized (sodium pentobarbital, 40 mg/kg, blocked the cats' capacity to recall the i.p.), and temperatures in cortex were measured with micro- discriminations (column C in i of Fig. 3 A-D), with perfor- thermocouples (Omega Engineering, Stamford, CT). From mance not reliably different from chance (50%) for any these measurements we estimated the extent of deactivated discrimination. In contrast, the cats' ability to recall the same cortex based on the observation that evoked activity is silenced discrimination with vPS cortex warm was excellent (>80%; by cooling of brain tissue to .20'C (9, 11, 12). The brain was column W in i of Fig. 3 A-D). Moreover, for all four discrim- then fixed, and horizontal sections were cut and stained for inations, there was an absence of an order effect when the Nissl substance or myelin. order of testing with vPS cortex warm or cold was reversed (not shown). Relearning and Learning. Highly familiar masked pattern. RESULTS vPS cortex is critical for the relearning of highly familiar or Anatomy. Midsagittal section of the optic chiasm was obvi- overtrained masked-pattern discriminations (Fig. 3A ii). With ous under oblique fiber-optic illumination, and section of all vPS cortex deactivated, in addition to being unable to recall the visual callosal fibers (9) was confirmed by gently separating the discrimination, both cats showed little evidence of being able two cerebral hemispheres. In addition, microscopic examina- to relearn the discrimination in 500 trials carried out over a tion of histological sections showed that the microstructure of 4-day period. During this period, performance fluctuated vPS cortex was unaffected by the: (i) surgical procedures, (ii) about chance and never exceeded 64% correct responses in a presence of the cortical cooling probes, or (iii) repeated single block of trials (Fig. 3A ii). In contrast, on day 5 with vPS coolings (Fig. 1). cortex active, performance was excellent and criterion perfor- Extent of Cortical Deactivation. We estimated the position mance was quickly achieved within 50 trials by both cats (e.g., of the 20°C thermocline by making >100 temperature mea- Fig. 3A iii). surements around the cooled probes in each cat. As can be Highly familiar simple pattern. Even though recall of highly seen in Fig. 1, temperatures of <20'C did not penetrate further familiar simple pattern discriminations was completely than 2 mm into cortex and neither encroached upon the white blocked by cooling of vPS cortex, the cooling did not prevent matter nor reached the hippocampus and its nearby structures. relearning of the same discrimination (Fig. 3B), which was From such multiple measurements, we reconstructed the relearned in 75-100 trials on day 1. (Fig. 3B ii). Even so, neither region of cortex deactivated by the cooling. In both cats, the cat showed evidence of savings of the relearned discrimination deactivated region encompassed a distance no greater than 2 on successive days with vPS cortex deactivated (Fig. 3B ii) mm from the probe and included all of the vPS cortex, much because initial performance on each day was close to chance. of the fusiform gyrus, and the posterior-most aspect of the However, with continued exposure to the discriminanda, the posterior ectosylvian gyrus (Fig. 2). This region corresponds to cats repeatedly relearned the discrimination in 75-125 trials. virtually all of area 20, ventral-most part of area 21, and part Recall on day 5, when vPS cortex was active, was excellent and of area EP (13, 14, 17), with little or no inclusion of parts of no relearning was needed (Fig. 3B iii). areas 17, 18, or 19 (Fig. 2). Any slight cooling of the white Highly Familiar Object. While recall of the highly-familiar matter would not have significantly altered action potential object discrimination was completely blocked during cooling Downloaded by guest on September 29, 2021 1656 Neurobiology: Lomber et al. Proc. Natl. Acad. Sci. USA 93 (1996) A(i)

CatT (I) CatSd i) Areas pig 17~~~~~~~~~~~~~~~~~~~~~ fg \~~~~~~~~~~~~~~~~~~~~~~~~20 1cm

FIG. 2. (A and B) Schematic illustrations of the cat cerebrum showing estimates of regions deactivated during cooling (hatching) and the positions of the cooling probes (wide black line within hatching) in Cat T (A) and Cat Sd (B). (C) Position of visual areas (13-15). (i) Lateral view (left is posterior). (ii) Posterior view (left is medial). fg, Fusiform gyrus; pesg, posterior ectosylvian gyrus; plg, posterolateral gyrus; rs, rhinal sulcus. Cat brain templates are adapted from the drawings of Reinoso-Suarez (16). deactivation of vPS cortex, the cooling did not severely impair pensations emerge following ablations, but not during cooling. relearning of the same discrimination (solid black I and 0 Such compensations likely involve peripheral or secondary cir- blocks; Fig. 3C ii), which was quickly relearned by both cats in cuits that contribute little to form discrimination in the normal 100-125 additional trials (Fig. 3C ii). Moreover, there was brain. These circuits are insufficient to either (i) permit imme- some savings of the relearned discrimination from day to day, diate access to an established mnemonic trace or (ii) be used as identified by sequentially higher performance on the first effectively for learning of new-object discriminations or relearn- block of trials on successive days. When subsequently tested ing of familiar masked-pattern discriminations during the limited with vPS cortex warm, both cats reached criterion levels of period vPS cortex is deactivated. However, these circuits may performance within 50 trials (Fig. 3C iii). contribute to the relearning of familiar simple-pattern and fa- Novel Object. vPS cortex is critical for the learning of novel miliar-object discriminations and the modest day-to-day savings object discriminations. During cooling deactivation of vPS of the latter discriminations. cortex, both cats failed to learn to distinguish between two Why might there be neural compensations following abla- previously unseen objects; even after 500 trials were per- tions? The dominant factor is likely to be the duration of the formed over 4 consecutive days (Fig. 3D ii). Even though deactivation. Following ablations of vPS cortex, cats live performance was above chance levels, it never exceeded 72% permanently with the neural defect and, because of interac- correct responses (Fig. 3D ii). On the next day, when vPS was tions with the environment, there is likely to be great pressure active, initial performance and rate of learning was in the for the cat's nervous system to compensate for the functional normal range and there was no evidence of prior exposure to deficits and reduce the handicap. In contrast, the pressure to the discriminanda (Fig. 3D iii). compensate for cooling-induced deficits is low because vPS cortex is deactivated for <10% of each day, and the brain functions normally for the remainder of each day. Mechanistic DISCUSSION factors that may contribute to the lesion-induced compensa- We have reached two general conclusions from our study of tions, which are not factors in cooling experiments, are alter- reversible cooling deactivation of vPS cortex: (i) vPS cortex is ations that the lesions induce at the molecular level by (i) essential for the recall of both object and pattern discrimina- triggering irritative responses by neurons or (ii) altering bal- tions, and (ii) vPS cortex is essential for both the learning of ances in trophic substance-neuron interactions. Both of these unfamiliar object discriminations and for the relearning of factors may lead to partial regenerative or sprouting responses highly familiar masked pattern discriminations. that serve to ameliorate the severity of the deficit. Deficit Specificity. We can be confident that the cooling- Cat vPS and Monkey IT Cortex: Are They Homologues? induced deficits are region and task specific because: (i) the Similar gross circuits involving the lateral geniculate nucleus, deactivated region is iocalized to within 2 mm of the cooling primary visual cortex, and transcortical relays, as well as probe; (ii) cooling of the middle suprasylvian region does not pathways through the superior colliculus, reach vPS cortex in impair recall of the same highly familiar simple- and masked- cats and IT cortex in monkeys (2). In both cats and monkeys, pattern discriminations (8, 18); and (iii) cooling of middle the most dominant cortical projections to vPS and IT cortices suprasylvian, but not vPS, cortex, impairs ability to discrimi- are derived from areas Vl, V2, and V3 (20-22). Furthermore, nate between different directions of motion (8). subcortical visual projections originating in the superior col- Comparison of Cooling- and Ablation-Induced Deficits. liculus reach vPS and IT cortices via the extrageniculate visual Several similarities and discrepancies exist between the cool- nuclei: lateral-posterior and pulvinar, respectively (20, 23-26). ing-induced deficits of this study and the ablation-induced These similarities suggest that the neural operations carried deficits of previous studies. While cooling and ablations ofvPS out by cat vPS and monkey IT cortices are broadly similar. This cortex both impair the learning ofvisual forms, cooling impairs suggestion is largely supported by behavioral studies when the learning of both relatively easy object discriminations and similar procedures are used to inactivate cortex (Fig. 4). In the more difficult patterns (5), whereas vPS ablations only both cats and monkeys, reversible cooling deactivation induces impair the learning of more difficult pattern discriminations deficits in learning and recall of form discriminations that are (3-5, 19). Another discrepancy is that vPS ablations have little as great as, if not greater than, lesion-induced deficits (Fig. 4; or no effect on the recall of previously learned discriminations 27-30, 34) when the regions inactivated by the two methods (3-5), whereas cooling impairs recall of all pattern and object are equated for size. The absence of an effect on cat recall discriminations. These discrepancies suggest that neural com- performance following ablations of vPS cortex suggests that Downloaded by guest on September 29, 2021 Neurobiology: Lomber et al. Proc. Natl. Acad. Sci. USA 93 (1996) 1657 Cat T Cat Sd A HIGHLY-FAMILIAR MASKED PATTERN Recall Relearning Discriminanda Recall Relearning 100- (i) (ii) (iii) (i) (ii) (iii) -100 90- I 0 90 0 I 80- (.)0 80 *c 70- 70 0 L- 60- 60 50 - I ", (chance) 50 1 2 4- 5 Day . 9 I 0 I kV1rk-13 4_ 5 Day - 11.1 I-- c w Cold Warm C W Cold Warm 25-Trial Blocks 25-Trial Blocks B

0U C)E U 0

u 0 25-Trial Blocks 25-Trial Blocks

1-U 0 IL

25-Trial Blocks 25-Trial Blocks D .' r %. .r Recall Learning Learning 100- (i) (ii) (i5i) (ii) -100

u 90- - 90 0 I o 80- - 80 oc U 70- - 70 10 0- 60- - 60

50- - (chance) I- _- - 50

1 2 4 I Day 1 2 3 4 5 Day _ 1 -~~~~~~~~~~~~~~~~~~. C W Cold Warm Cold Warm 25-Trial Blocks 25-Trial Blocks

FIG. 3. (Left and Right) Learning and recall of form discriminations for Cat T (Left) and Cat Sd (Right). (A, B and C) Data on recall and relearning of highly familiar masked pattern (A), simple pattern (B), and object (C) discriminations. (D) Data on the learning of a novel object discrimination and its subsequent recall. (Center) The pairs of positive (rewarded) and negative (unrewarded) discriminanda between data sets for the two cats. Each data set has three components: (i) recall of the pattern or object pair while vPS cortex was cold (column C; dark stipple) or warm (column W; light stipple) with standard errors indicated where applicable; (ii) (re)learning of the discrimination while vPS cortex was cold (0) on 4 consecutive days; (iii) (re)learning of the same discrimination while vPS cortex was warm (0) on day 5. For uniformity, object recall (D i) appears before object learning (D iii), although in reality, this order was reversed. Circles = 25-trial blocks. Chance = 50%. Downloaded by guest on September 29, 2021 1658 Neurobiology: Lomber et al. Proc. Natl. Acad. Sci. USA 93 (1996)

COMPARATIVE DEFICITS IN FORM DISCRIMINATION 9. Lomber, S. G., Payne, B. R. & Rosenquist, A. C. (1994) Behav. Brain Res. 64, 25-35. 10. Payne, B. R., Pearson, H. E. & Berman, N. (1984) J. Neuro- Learning Recall physio. 52, 570-594. 11. Benita, M. & Conde, H. (1972) Brain Res. 36, 133-151. a ~~a Datvto CAT la llReversible 12. Jasper, H. H., Shacter, D. G. & Montplaisir, J. (1970) Can. J. Physio. Pharmacol. 48, 640-652. CAT I l Deactivation 13. Updyke, B. V. (1986) J. Comp. Neurol. 246, 265-280. vPS Cortex c g 14. Bowman, E. A. & Olson, C. R. (1988) J. Comp. Neurol. 272, Permanent 30-42. ______,tO,, ~PtAblation 15. Tusa, R. J., Palmer, L. A. & Rosenquist, A. C. (1981) in Cortical Sensory Organization; Multiple Visual Areas, ed. Woolsey, C. N. 11_ 11 ~~~~~Reversible (Humana, Clifton, NJ), Vol. 2, pp. 1-31. 16. Reinoso-Suarez, F. (1961) TopographischerHirnatlas derKatzfur MONKEY Deactivation experimentalephysiologische Untersuchungen (Merck, Darmstadt, Germany). ITCotex to\ to / Ablation 17. Tusa, R. J. & Palmer, L. A. (1980) J. Comp. Neurol. 193, 147-164. 18. Lomber, S. G., Cornwell, P., Sun, J.-S., MacNeil, M. A. & Payne, B. R. (1994) Proc. Natl. Acad. Sci. (USA) 91, 2999-3003. 19. Hughes, H. C. (1982) in Analysis of Visual Behavior, eds. Ingle, FIG. 4. Key: I l = severe deficit; = deficit; 0 = no deficit. D. J., Goodale, M. A. & Mansfield, R. J. W. (MIT Press, Cam- Superscript lowercase letters refer to references: a, present study; b, bridge, MA), pp. 771-800. ref. 4; c, refs. 3 and 5; d, ref. 27; e, refs. 28-30; f, refs. 31 and 32; and 20. Cavada, C. & Reinoso-Suarez, F. (1983) Brain Res. 270, 319-324. g, refs. 1 and 33. 21. Symonds, L. L. & Rosenquist, A. C. (1984) J. Comp. Neurol. 229, 1-38. the ablations may have been less than complete. The broad 22. Baizer, J. S., Ungerleider, L. G. & Desimone, R. (1991) J. similarities in circuitry and behavioral deficits induced by Neurosci. 11, 168-190. reversible cooling deactivation in the two species suggest that 23. Benevento, L. A. & Standage, G. P. (1983) J. Comp. Neurol. 217, the neural operations carried out by cat vPS and monkey IT 307-336. cortex are largely similar. 24. Abramson, B. P. & Chalupa, L. M. (1988) J. Comp. Neurol. 271, 397-418. 1. Gross, C. G. (1973) in Handbook ofSensory Physiology, ed. Jung, 25. Baleydier, C. & Morel, A. (1992) Vis. Neurosci. 8, 391-405. R. (Springer, Berlin), pp. 451-482. 26. Baizer, J. S., Desimone, R. & Ungerleider, L. G. (1993) Vis. 2. Payne, B. R. (1993) Cereb. Cortex 3, 1-25. Neurosci. 10, 59-72. 3. Sprague, J. M., Levy, J., DiBerardino, A. & Berlucchi, G. (1977) 27. Horel, J. A., Voytko, M. L. & Salsbury, K. (1984) Behav. Neu- J. Comp. Neurol. 172, 441-488. rosci. 98, 310-324. 4. Campbell, A. (1978) J. Comp. Physiol. Psychol. 92, 45-57. 28. Horel, J. A., Pytko-Joiner, D. E., Voytko, M. L. & Salsbury, K. 5. Cornwell, P. & Warren, J. M. (1981)J. Comp. Physio. Psychol. 95, (1987) Behav. Brain Res. 23, 29-42. 603-614. 29. Cirillo, R. A., George, P. J., Horel, J. A. & Martin-Elkins, C. 6. Horel, J. A. (1991) in Lesions and Transplantation: Methods in (1989) Behav. Brain Res. 34, 43-53. Neurosciences, ed. Conn, P. M. (Academic, San Diego), Vol. 7, 30. Horel, J. A. (1992) Behav. Brain Res. 51, 193-201. pp. 97-110. 31. Ando, Y. (1969) Nagoya J. Med. Sci. 31, 485-507. 7. Cornwell, P., Herbein, S., Corso, C., Eskew, R., Warren, J. M. & 32. Cowey, A. & Gross, C. G. (1970) Etp. Brain Res. 11, 128-144. Payne, B. (1989) Behav. Neurosci. 103, 1176-1190. 33. Iwai, E. & Mishkin, M. (1969) Exp. NeuroL 25, 585-594. 8. Lomber, S. G., Payne, B. R., Cornwell, P. & Long, K. D. (1996) 34. Fuster, J. M., Bauer, R. H. & Jervey, J. P. (1981) Exp. Neurol. 71, Cereb. Cortex, in press. 398-409. Downloaded by guest on September 29, 2021