Functional interactions between inferotemporal and prefrontal cortex in a cognitive task

Brain Research, 330 (1985) 299-307 Elsevier

299

BRE 10625

Functional Interactions Between Inferotemporal and Prefrontal Cortex in a Cognitive Task JOAQUIN M. FUSTER, RICHARD H. BAUER and JOHN P. JERVEY

Department of Psychiatry and Brain Research Institute, School of Medicine, University of California at Los Angeles, Los Angeles, CA 90024 (U.S.A.) (Accepted July 5th, 1984)

Key words: inferotemporal cortex - - prefrontal cortex - - cooling-- units - - vision - - short-term memory-- monkeys

Monkeys were trained to perform a visual short-term memory task (delayed matching to sample). In some of the animals, cooling probes were implanted over dorsolateral prefrontal cortex, covering sulcus principalis and adjacent areas; microelectrode pedestals were implanted over inferotemporal cortex. Other animals were fitted with converse implants: cooling probes over a portion of the inferotemporal cortical convexity and microelectrode pedestals over prefrontal cortex. In the awake and behaving monkeys, bilateral cooling of either the prefrontal or the inferotemporal region (to 20 °C) induced, in the other region, reversible changes of spontaneous and task-related cell discharge. In the two cortices remote cooling induced augmentations and diminutions of cell reaction to the color samples which the animal had to retain for correct performance of the task. The same was true for cell discharge during the delay, the retention period which followed each sample. However, a net effect of remote cooling was, in both cortices, a diminution of color-dependent differences in the reactions and delay-discharge of some cells. Concomitantly, errors of task-performance increased. Cells that as a result of remote cortical cooling showed changes of reaction to the color samples were found more commonly in supragranular than infragranular layers. The results are interpreted as evidence of mutual influences between inferotemporal and prefrontal areas, probably mediated by corticocortical connections. The single-cell data, together with the behavioral data, suggest that those influences are functionally important for visual discrimination and short-term memory. INTRODUCTION In the m o n k e y , visual short-term m e m o r y is impaired by the functional depression, by cooling, of either prefrontal cortext, 8 o r i n f e r o t e m p o r a l cortex 10, though not parietal cortexL Thus it seems that both prefrontal and i n f e r o t e m p o r a l areas participate in short-term retention of visual data. It m a y be inferred that those two widely s e p a r a t e d cortices are part of a widely distributed system of visual representation and that both are activated by behaviorally relevant visual stimuli for as long as the situation demands it. F u r t h e r evidence of that is p r o v i d e d by single-unit studies. Both prefrontal7,11,19 and inferotemporall2,17 cells react to visual stimuli in the context of short-term m e m o r y tasks. Some of those cells - - unlike parietal cells 7 - show sustained and stimulus-dep e n d e n t changes of discharge after a stimulus has disa p p e a r e d and for as long as the animal must retain it to meet the task's requirements7a 2.

Because of the m e n t i o n e d similarities between prefrontal and i n f e r o t e m p o r a l cortex with regard to cooling effects as well as cellular reactivity, and because of the reciprocal connectivity b e t w e e n the two corticesaAaA6,20, 21 it seems reasonable to postulate a close functional relationship b e t w e e n t h e m in visual discrimination and short-term m e m o r y . The present study examines that relationship. W e test the effects of reversible lesion of one cortex on unit activity in the other, as well as on behavior, in the course of performance of a visual short-term m e m o r y task. A preliminary r e p o r t was p r e s e n t e d elsewhere ~3. MATERIALS AND METHODS

Subjects Nine male m a c a q u e monkeys, weighing about 8 kg, were used in these experiments. Six of t h e m were also used for a study of i n f e r o t e m p o r a l unit activity te and the o t h e r 3 for a study of prefrontal unit activ-

Correspondence: J. M. Fuster, UCLA Neuropsychiatric Institute, 760 Westwood Plaza, Los Angeles, CA 90024, U.S.A. 0006-8993/85/$03.30 © 1985 Elsevier Science Publishers B.V. (Biomedical Division)

300 ity n. They were housed in independent cages and maintained on ad libitum diet.

Behavior The experimental animals were fully trained to perform a delayed matching-to-sample (DMS) task with colored stimuli. The subject sat in a primate chair in front of a white panel with translucid stimulus-response buttons (2.5 cm in diameter). A trial started with the appearance of the sample, a colored light, in a centrally located button. The subject turned it off by pressing the button. A delay (16-20 s) followed, at the end of which two or 4 colors, one of them the sample color, appeared simultaneously on a row of buttons in the lower part of the panel. If the monkey then pressed the button with the sample color it received a squirt (about 1 ml) of fruit juice through a metal spigot. The choice of any other color terminated the trial without reward. Both the color of the sample and its position in the choice buttons were changed quasirandomly from one trial to the next. Therefore, for correct performance of the task, the animal had to remember the color of the sample for each trial without the aid of spatial cues. Four of the experimental animals performed the task with two colors, red and green, as sample and choice colors; the other 5 animals performed with red, green, blue and yellow. The buttons were transilluminated by rear projection of white light through Cinemoid color filters. The spectral characteristics and intensity of the stimuli were determined and adjusted with a Pritchard Spectra photometer. All stimuli subtended 8° of visual field. The sample had a luminance of 13.5 cd/m 2 (+ 0.1 log unit). Dominant 2 was 620 nm for red, 590 nm for yellow, 530 nm for green and 480 nm for blue.

Surgery Surgery was carried out with the animal under general Nembutal anesthesia and with the help of stereotaxic guidance. In one group of 6 animals, cooling probes and thermistors were bilaterally implanted over dorsolateral prefrontal cortex. The round, goldplated, copper probes were of the same type and covered approximately the same cortical region as those used for other studies of prefrontal cooling 1. That region included a large portion of area FD - - as defined by von Bonin and Bailey 3 - - bisected by the sulcus

principalis. In addition, hollow pedestals for a microelectrode positioner were implanted over inferotemporal cortex (area TE). In a second group, of 3 animals, the reverse surgical procedure was carried out. Elongated cooling probes and thermistors were bilaterally implanted on the convexity of the inferotemporal cortex 10 and micropositioner pedestals over dorsolateral prefrontal cortex. All implants were affixed to the skull by means of stainless steel screws and acrylic cement. Threaded metal sleeves for eventual fixation of the head during recording were imbedded in the cement mound.

Experimentalprocedure In preparation for an experimental session the animal was placed in the testing apparatus and thermoelectric coolers, operating on the basis of the Peltier principle, were attached to the two prefrontal or inferotemporal probes. A hydraulic positioner 6 with a tungsten or platinum-iridium microelectrode (0.5-3.0 Mr2) was mounted on either an inferotemporal or a prefrontal pedestal, as the case might be (inferotemporal pedestal on animals with prefrontal coolers and prefrontal pedestal on animals with inferotemporal coolers). The potentials recorded with the microelectrode were amplified (capacitycoupling) and monitored with an oscilloscope. The microelectrode was advanced slowly through the cortex while the animal was tested on DMS with short delay (6-10 s). The extracellularly recorded action potentials from single cells were converted into standard pulses (0.5 ms, 1 V) by means of a Schmitt trigger and recorded on magnetic tape with an Ampex FR-1200 recorder. A digital code and DC pulses marking the trial events were recorded on a separate channel. Upon isolation of a unit and verification of the stability of its record, the animal was submitted to a series of DMS trials with delays between 16 and 20 s at intervals of about 50 s. The discharge of the inferotemporal and prefrontal units recorded under these conditions was the subject of two previous publications TM. A subset of those units, selected for testing the effects of cortical cooling, constitutes the subject of the present study. Each of those units was selected on the basis of the stability of the record and the presence of clear firing frequency changes in relation to the events in DMS trials. After recording the activity

301 of the unit through a number of trials, the coolers were activated and thus the underlying cortex was cooled, bilaterally, to 20 °C and thermostatically maintained at that temperature. The animal was then submitted to a second series of trials under cortical cooling. At the end of that series, the temperature of the cortex was brought back to normality and, if the unit record continued to show stability and good isolation, a third series of trials was administered. On occasional units, the cooling and rewarm procedures were repeated once or twice. Fig. 1 represents schematically the cooler locations and cortical regions explored with microelectrodes in the two groups of experimental animals.

Histology After completion of experiments with a given animal, electrolytic marks were made in the brain by passing small amounts of DC (100pA, 15 s) through a microelectrode. Such tissue marks allowed later reconstruction of unit locations. The animal was sacrificed with an overdose of Nembutal and the brain fixed in formalin and cut in coronal sections (80 ~m). The estimated position of all the recorded units was marked on photographic enlargements of the Nisslstained sections.

~P ~

STS

Fig. 1. In 3 of the subjects, inferotemporal cortex was bilaterally cooled (shaded area, above, left) and single-unit activity recorded from a portion of prefrontal cortex (marked below in cross-section). In 6 other subjects, prefrontal cortex was bilaterally cooled (shaded area, above, right) and single-unit activity recorded from inferotemporal cortex. The position of the arrows with respect to the upper diagrams indicates the planes of the cross-sections underneath. SP, sulcus principalis; STS, superior temporal sulcus.

Analysis The analysis of cell discharge was done off-line by means of a MINC-11 computer. Spike trains were digitized for all the available behavioral trials during the recording of any given unit. Each trial-series, whether at normal cortical temperature or during cooling, consisted ordinarily of 6-12 trials with each of the sample colors. Frequency histograms of the unit's activity, time-locked with the sample, were obtained for each sample color both with and without cortical cooling. The firing frequency changes of each unit after sample presentation were statistically analyzed, using firing frequency in the 16 s preceding each trial as the base line. The entire epoch of the trial was divided into time bins. Beginning with sample onset, the unit's frequency within each bin was subtracted from base-line frequency. The resulting difference scores for each bin after sample onset were averaged across all trials of the same sample and condition (precooling, cooling, or rewarm). The mean and standard deviation of those differences scores were used in a t test for correlated means, to ascertain whether within that time bin the unit's probability of firing deviated from base line. Thus, the unit's deviations from base-line firing, during the sample period as well as the subsequent delay, were analyzed throughout the entire series of trials for each sample color and condition. The use of difference scores, as described, for comparing intratrial firing with preceding base-line firing, allowed the assessment of unit reactivity to the color samples independently of fluctuations in baseline firing over time, whether those fluctuations occurred spontaneously or as a result of cooling. Comparisons (per bin) between sample colors were done with t tests of the differences between the difference scores for each color. The same procedure was followed for comparing conditions (precooling, cooling, rewarm), that is, for testing the effect of cooling on the activity of a unit during and after samples of a given color. All the units of this study were tested in precooling and cooling conditions, as the stability and isolation of their record could be assured throughout the time (about 1 h) that behavioral testing took in those two .conditions. Reversibility of cooling effects (by comparing rewarm with precooling and cooling data)

302 TABLE I Effects of inferotemporal cooling on prefrontal units (n = 49) in delayed matching to sample Percentages are given in parentheses. Spontaneous discharge Sample reaction Unchanged Increased Decreased

30 (61) 10 (21) 9 (18)

37 (76) 7 (14) 5 (10)

could only be tested in those units (36%) that could be recorded through at least one complete cycle that included the 3 conditions (about 2 h). In all statistical tests, only P values less than 0.5 were considered to denote statistically significant differences. RESULTS General properties of the units investigated Practically all the 49 prefrontal units of this study were situated in lateral and ventral aspects of prefrontal cortex, that is, in the cortex of the lower bank of the sulcus principalis, the inferior prefrontal convexity, and the orbital surface. Three of the units were situated in the upper bank of that sulcus. All the 59 inferotemporal units were situated in the cortex of the lower bank of the superior temporal sulcus and the inferotemporal convexity. The average spontaneous discharge of the prefrontal units was relatively low (mean, 4.6 spikes/s; S.D., 6.6; median, 2.3 spikes/s). That of the inferotemporal units was higher (mean, 6.5 spikes/s; S.D., 7.4; median, 3.4 spikes/s). As previously reported n,12, a substantial proportion of cells in the cortical areas explored for this study show excitatory reactions to the sample stimulus in the DMS task. That proportion is larger in inferotemporal than in prefrontal cortex. Also more

Delay activity 33 (67) 10 (21) 6 (12)

Color differentiation Sample

Delay

41 (84) 0 (0) 8 (16)

45 (92) 0 (0) 4 (8)

common in inferotemporal cortex are cells that react differentially to the sample colors. In general terms, cell reactions to the sample colors are brisker and more apt to be color-differential in inferotemporal than prefrontal cortex. Many of the units investigated, in prefrontal as well as inferotemporal cortex, showed sustained activation or inhibition during the delay period of the DMS task. Some units showed color-dependent ac-. tivity during that period (though the color sample was no longer present). Again, that kind of differential activity was more common, and the color-dependent differences more marked, in inferotemporal than prefrontal cortex. Effects of cortical cooling Tables I and II summarize changes induced by inferotemporal and prefrontal cooling on, respectively, prefrontal and inferotemporal units. In either region, remote cortical cooling affected the spontaneous activity and the task-related activity of substantial numbers of ceils. Spontaneous activity. Remote cooling increased the spontaneous discharge of some units while decreasing that of others. The direction of change was not clearly related to either the location or the functional properties of the units examined. Increases of spontaneous firing were seen more often than de-

TABLE II Effects of prefrontal cooling on inferotemporal units (n = 59) in delayed matching to sample Percentages are given in parentheses. Spontaneous discharge Sample reaction Unchanged Increased Decreased

27 (46) 19 (32) 13 (22)

43 (73) 7 (12) 9 (15)

Delay activity 38 (64) 5 (8) 16 (27)

Color differentiation Sample

Delay

50 (85) 0 (0) 9 (15)

56 (95) 0 (0) 3 (5)

303 NORMAL

iillllllllllll

i

IIIII

IIIIIIIIII

, "£"llll',.I I Ill

~t

II

l~il

I I

Ira"

II IIIIIII I

I HIIII~ II11 I l i l

I

t J

III

II

,

I I

I

[I

IIIIIIIII

HIIIIIIII

J

t

JJ I

i

II

IIIIIIII

lib

I1

IIIIIIII

I

III

I II "~1111 I I I II I IIl'i~l I I I I

lit

I

ill

I

i lll~'i

III

I, I llllll

II

II III II

-IlPllll

II

IIIIIIII

III

III1 III

0

5

10

-5

0

5

10

TIME

-5

I

IIIIIIllllll

I I

I II

III 11 1

I I

lill,nlllLi

IIIII

II

ilHIII

JUllllJllll.

11111111111

I111111 mn

i i!11tll I

i IIIIIIIIIlllllll

I IIIIII

R

-5

I11

I II

1111 II

Illl

Itl

mmHnlnilJulnJ

Ilill

I

III

Illlllllllllllllllllllll i,l~

pl_._nlilllllll

.n

I II

III1

I I II IIIIIIIII

]illllll

__11

I Ill I

i

I

lilil

I

1

IfllllllilHIII

Iltll I Ill

I

IIII IIIg

iililill

I'i'~

I IIIIIII-BI

IIIIIIIII I II II I

.

il~'Ol t J

"~'t

II111 I i IIIIII/li

II

,71

I1 I Ill

II II

IT C O O L

I

mlmlil II

I111 Illllll

G

0

5

10

-5

0

5

10

(SEC)

Fig. 2. Spike reactions of a prefrontal cell to the sample colors of DMS trials at normal cortical temperature and during bilateral cooling (20 °C) of inferotemporal cortex (IT COOL). Rasters of spike discharge show cell reactions to the sample (horizontal bars) in individual trials (R, red; G, green). The average frequency histograms are shown below.

creases, especially in inferotemporal cortex (under prefrontal cooling). Reaction to the sample. As a result of remote cooling the magnitude of the reaction of many cells to the sample stimulus was changed; in some cases it was augmented and in others it was diminished. However, cooling did not necessarily have the same effect on the responses of individual cells to various sample colors. In fact, the color-dependent differences of reaction that some cells exhibited, in both prefrontal and inferotemporal cortex, were diminished or abolished (in no case augmented) by remote cooling; as a result of cooling the preference of a cell for a given color frequently disappeared (Figs. 2 and 3). In other words, under c o o l i n g the cells had a t e n d e n c y to treat

different colors alike and to abandon their normal color preference if they had one. At the same time,

NORMAL

120

--

Red*

COOLING

--

Yellow

(61% C O R R E C T )

100

Ez

8o ~- Bo ~ 40 ~ 20 < O ~ -20z~ -40~ -60rr tu -80 o_ -lOO-

Yellow Bed

I

Green Green * Blue*

Blue *

15"14A (85%CORRECT)

(76% CORRECT)

200 180 Z rr

160 -

--

Green *

--

Red *

--

Green *

140 -

~

120 100 -

Fig. 3. Average firing changes (in percent of pretrial baseline frequency) induced by 4 color samples in the DMS task on two cells in inferotemporal cortex (top two graphs for one cell and bottom two for the other). On the left graphs, color-dependent changes in normal condition; on the right, color-dependent changes under bilateral prefrontal (PF) cooling (20 °C). Percentages of correct behavioral responses of the animal in the task are given in parentheses. Statistically significant (P < 0.05) changes from baseline-firing are indicated by an as-

PF

(78% CORRECT)

z < "r" o

60-

~z

4 0 ~.

--

o 07

20-

---

ILl

--

Blue *

--

Yellow*

8O-

Blue Red Yellow

-20 15-43C

SAMPLE

SAMPLE

terisk. In both cells cooling diminishes differentiation of the colors (top cell normally favors red, bottom cell green), while correct performance is also diminished.

304 NORMAL

O.

PF COOL

s

c

s

£

¢,n

REWARM

,o 1

10

5

0

5 TIME

10

15

20

25

(SEC)

P121 608

Fig. 4. Average firing frequency of a cell in inferotemporal cortex during 3 blocks of trials in the DMS task. Top, normal temperature; middle, prefrontal cortex bilaterally cooled to 20 °C; bottom, cortex rewarmed to normal temperature. S, sample period; C, choice period. Note that prefrontal cooling attenuates the normal activation of firing during the delay period.

No clear patterns were observed in the topographic distribution of the units whose spontaneous discharge was changed, one way or the other, by remote cooling. They seemed to be represented about equally in all cortical layers. That was not the case, however, for the reaction to the sample and the subsequent delay activity of the units investigated. When the sample-induced changes of firing were considered, a difference in prevalence of units affected by cooling was noted between layers above and layers below the internal granular layer (IV). (Only units which could be located with confidence either above or below layer IV were included in the comparison; units presumed to be in that layer were not included because their histological localization did not allow the same degree of confidence.) In both prefrontal and inferotemporal cortex, units whose reactions to

NORMAL

the animal committed more errors of DMS performance. Activity during the delay. The discharge of cells during the delay period was also modified by cooling. In both cortices remote cooling depressed the delay activity of some cells and excited that of others. Enhancement of delay discharge was more common than depression in prefrontal cortex, whereas the reverse was true in inferotemporal cortex (g 2 = 5.62; P < 0.02) (Fig. 4). As in the case of the immediate reaction to the sample colors, color-dependent differences in activity during the delay were diminished or obliterated by remote cooling (Figs. 5 and 6). Concomitantly, the animal's performance became less accurate. Reaction to the choice lights. The visual stimulation at the end of each DMS trial, when two or 4 colored lights were presented for choice, was obviously quite complex. Therefore, unit-activity changes occurring at that time were difficult to evaluate and no attempt was made to analyze in standard fashion the reactions of all the units to the choice lights. However, certain units showed clear-cut changes of those reactions as a result of remote cooling. Under inferotemporal cooling, a few prefrontal units exhibited remarkable increases of firing at the time of choice (Fig. 7).

Laminar topography of units affected by cooling.

IT COOLING

(89% CORRECT

(59%CORRECT)

300 z

250'

t7

200

w (5 z < I

100

]

150

o ~

50

O

-

-

Re d *

l / iI Green

i /

k"

/

/

/

_ _ R e d . +.'~"--~'Green*

-50 -100

Pr

LU - 1 5 0 O_ -200 12-8c (100% C O R R E C T

(87% C O R R E C T )

30 (5 z

25 20

t

Green*

/

~

10

z < I

5

~ z w o ~

I / I

0 - -

- - R e d

-5

-,-- \

-

\

-tO

Red

Green

-15 -20 DELAY

DELAY

20-23A

Fig. 5. Average firing changes (in percent of pretrial baseline .frequency) induced by two color samples on two prefrontal cells (upper graphs for one cell, lower graphs for the other) during the delay period of the DMS task. At left, color-dependent changes in normal condition and, at right, under bilateral inferotemporal (IT) cooling (20 °C). Notations on behavioral performance and on statistical significance of firing changes, as in Fig. 3.

305 NORMAL

PF C O O L I N G

(100% CORRECT

z

E

301 25 20

f

-

-

(59% CORRECT)

Green *

I

z<

z

5

I I // r

Green*

/ Red

0

t/.-

' Red

Z

w -10 ~ -15 w -20 7-3D

(93% CORRECT)

(68%CORRECT)

30 z

25

~

20

Red* I i t I

z< ~

5 0

~ Z

-5

L

\~

Yellow Blue

\

w -10 w

Green

/t

I

Green

( -

Yellow

\

Red Blue*

\x

-15

-20 15"31B

DELAY

interpretation of the data, it is reasonable to assume that hypothermic cortex was to some degree, and reversibly, depressed is. Thus the cellular changes induced by remote cortical cooling are attributable to the temporary subtraction o f - direct or indirect - i n p u t from the area cooled and are not likely the expression of some form of denervation suprasensitivity. Following the ablation of a cerebral structure, suprasensitivity may develop elsewhere, but only after periods that are considerably longer (over 24 h) than our cooling periods, which usually lasted between 30 and 60 min 2,23. Our data indicate that the removal or depression of function of either of the two cortical regions investigated leads to both excitation and inhibition of cell discharge in the other. Thus, excitation as well as inhibition are probable net effects, at the cell level, of the mutual influxes that the two regions normally exchange. The functional importance of both excitatory and inhibitory phenomena may be inferred from the finding of both increases and decreases of firing

DELAY

Fig. 6. Average firing changes (in percent of pretrial baseline frequency) induced by color samples on two cells in inferotemporal cortex during the delay of the DMS task. At left, normal temperature; at right, bilateral prefrontal (PF) cooling. Notations on performance and statistical significance, as in Fig. 3. the sample were affected by remote cooling were significantly more c o m m o n in supragranular layers than in infragranular layers (Fig. 8). DISCUSSION It may be inferred in general terms that the two cortical areas explored, prefrontal and inferotemporal, are normally under the influence of each other and thus the cooling of either area alters the influence of that area on the other. However, it cannot be assumed that the cortex underlying a cooling probe was uniformly cooled. Nor can it be assumed that hypothermia was circumscribed to the cortical patch immediately below the probe, for it has been shown that - - with the surface at 20 °C - - the temperature gradients around a cooling probe, though rather steep, are not sharp9. Furthermore, the prefrontal inferotemporal areas cooled did not completely coincide with the areas in which units were sampled. Despite those methodological constraints on the

SAMPLE

CHOICE S

R

, L : .TCOOL! ,., REWA MI ,.' 25

o

25

25

0

2.5

25

0

2.5

TIME ( S E C ) PF2038B

Fig. 7. Firing frequency of a prefrontal cell at the sample and choice periods of the DMS task in 3 conditions: normal cortical temperature (top), bilateral inferotemporal cooling (middle), and after rewarming (bottom). Sample-period histograms are time-locked with the onset of the stimulus. Choice-period histograms are time-locked with the onset of the choice lights (S) and with the behavioral response (R).

306

REACTIVITY TO SAMPLES AFFECTED BY COOLING ? PREFRONTAL COOLING

INFEROTEMPORAL CELLS Yes ~

Supragranular

~

|

----Infragranular

NO

16 3

15 N:35 X~:18.13 p