Visual responses to reward-related cues in inferior parietal lobule W. A. MacKay, B. Blum* and A. J. Mendonga Department of Physiology. University of Toronto. Toronto M5S IAS, Canada (Received 8 October 1991) A sample of 263 neurones was recorded in area 7a of the parietal lobe, iti a monkey performing a reach task to visual targets displayed on a toueh-sensitive videomonitor. The task had been operantly conditioned on food or juice rewards, and 78 (30%) of the units showed activity changes linked in some way to the reward. For most of these cells, the response was to the approach of the trainer's hand with the food reward. This specific visual response was similar irrespective of the direction of approach. Six cells increased discharge as soon as the task was completed in apparent anticipation of the reward. Another two neurones responded to missing a reward: they fired vigorously if the videoscreen was blanked in mid-trial because a target was not correctly touched. In many cases (40/78) the same cells responding to some aspect of the reward also responded to visual cues given during the task, especially the presentation of the target location. Reward-related activity in area 7a probably results from an integration of the visual and limbic inputs to this region, such that visual information which foretells behaviourally important events is emphasized.

Although the inferior parietal lobule (area 7a in particular) is generally associated with the visual analysis of extrapersonal spatial relationships', it is known to receive major input from iimbic regions of the cerebral cortex^'^. Consequently, one would expect to find motivational, affective or appetitive factors represented there in some way. In the typical operant conditioning paradigm of behavioural neurophysiological studies, the reward provides a potent hmbic stimulus. In addition, many experimental protocols have built-in cues which reveal whether or not a reward is to be delivered. While recording in area 7a of a monkey performing a visually-guided reach task, we encountered many neurones which responded specifically to the delivery of a food or juice reward, or. in rare cases, to visual signals on the videomonitor which indicated whether or not a reward was forthcoming. Methods Experiments were performed in an adult, male monkey {Macaca fascieularis). weighing 5 kg. The monkey sat in a modified primate chair placed in front of a 20-inch colour videomonitor. The head was partially restrained in that it could not be translated in space, but was free to rotate rightward or leftward. The rotation of the head was monitored by a potentiometer coupled to a cable mounted within the restraining device. The horizontal EOG was recorded by means of paediatric ECG electrodes applied on each side of the head, lateral to the Correspondence lo: Dr W. A. MacKay • O n sabbatical leave from Ihe Department of Physiology and Pharmacology, Sackler School of Medicine, Tel Aviv University, Tel Aviv 69978, Israel.

© 1992 Butterworth-Heinemann for British College of Optometrists 0275-5408/92/020209-06

eyes. Eye position could be calculated from the EOG records, and by adding it to the head position, horizontal gaze angle was computed. Sample gaze records during task performance arc given in Figure I A. To initiate a trial, the monkey first positioned his arms on rest-plates situated close to his sides. Then the videoscreen illuminated with a coloured background to signal which arm to use, violet for the right arm or blue for the left. As shown in Figure IB. the target location was indicated 600 ms later by the appearance of an open square anywhere on the screen. The monkey could not move, however, until the end of a 1 s delay period cued by a GO-signal (the open square was replaced by a solid one). The monkey then reached with the appropriate arm to touch the target. Contact time and location on the screen were registered by means of an infra-red grid (Smart Frame, Carroll Touch, Round Rock. TX, USA) mounted in front of the videomonitor. If the target was correctly contacted (allowable window was 3 cm"), then the screen turned a light green colour accompanied by randomly generated tone sequences. The monkey was subsequently given a food (nut/raisin/sunfiower seeds) or fruit juice reward. All rewards were manually delivered to the mouth by the monkey's trainer. If targets were not adequately attained, the screen blanked and a new trial had to be initiated. Also, a second protocol was often used in which two targets had to be touched in sequence. After the first target was contacted no green light appeared. Instead, after a delay of 450 ms, a second target location was indicated by an open square, followed 1 s later by a solid square which the monkey then touched to turn the screen green and receive his reward. A recording well was implanted on the skull under general inhalation anaesthetic (halothane and NO^) and

Ophthal. Physiol. Opt., 1992. Vol. 12, April

209

Reward responses in parietal cortex: W. A. MacKay et al

75 -\ right

I. O)

(D

cn c N

D O

OFF HAND

TARGET

GO

TOUCH

Figure t The monkey's visually-guided reach task. (.4) A selected sample of gaze records collected dunng lask performance. (B) Schematie outlines of the videomonitor depict ihe sequence of visual cues which guide the task

aseptic conditions. A post-operative analgesic was administered (0.4 mg of buprenorphine HCI) and daily injections given of antibiotic (getitomycin) for 4 d. The scalp wound edges were treated with polysporiti ointment or nitrofurantoin spray. After 10 d, microeiectrode recording was started in the inferior parietal lobule of the left hemisphere, using standard chronic recording techniques. The electrodes were made of glass-coated Pt Ir wire with tip impedances of 0.5 to 2.5 Mil at 1000 Hz. All procedures were in accordance with the 'Guidelines for the Use of Animals in Neuroscience Research" published by the Society for Neuroscienee, USA. and were approved by a local animal care committee. The monkey was never deprived of water or food in his home cage. Results A sample of 263 neurones was recorded in area 7a as the monkey performed the task. Of these. 72 displayed activity directly related to the reward. As shown by the distribution of solid circles in Figure 2, reward-related cells were located throughout the explored territory of area 7a, which was restricted to the top of the gyrus. They responded to the approach of the trainer's hand with the food or juice reward. The example illustrated in Figure 3 shows how strong the reward response could be, with discharge rates in exeess of 150 impulses s ~ ^ as the reward approached, in spite of the virtual silence throughout the task. The direetion of reward approach was not discriminated by these neurones. If a reward was

210

Ophthal. Physiol. Opt., 1992. Vol. 12. April

Figure 2 Sites of electrode tracks in which reward-reiaied units were recorded. The shaded area in the inset diagram shows ihe location of area 7a in the cerebral cortex. Tracks are indicated by solid circles for responses to the reward itself (large solid circle locates site of unit in Figure 3]. An empty circle marks the site of the unil shown in Figure 4. and an empty square marks the site of the unit shown in Figure 5: ip, intraparietal sulcus

delivered either from the right side (Figure 3A) or from the left side {Figure 3B), the same discharge rates were elicited. In many cases (39/72)., the same cells responding to the approach of the reward also responded to visual cues given during the task, especially the presentation of the target location.

Reward responses in parietal cortex: W. A. MacKay et al.

IMS iiinraiii iiiiiiiiiiiiii I I •Mmiiiiiiiii II

I

REWARD

B

HAND

TARG

GO

GREEN ON

1

sec

Figure 3 Example of an area 7a neurone responding to the approach of a reward after successful completion of ihe task. | A) Reward was delivered from the right side. Rasters of the neuronal impulses are displayed for individual trials. (B) Reward was delivered from the left side

Two neurones responded to missing a reward, in addition to reward approach and to target presentation. Data from one of them, collected during performance of the two-target task, are illustrated in Figure 4. In sueeessful (i.e. rewarded) trials, a prominent burst of activity was observed in response to the appearance of the second target, a very weak response to the green field and a weak increase in discharge as the reward was presented (Figure 4A ). For those trials in whieh the first touch was inaccurate, the screen blanked a fixed interval after the appearance of the second target [Figure 4b). The blanking of the screen elicited a very vigorous burst of impulses from the neurone. Similarly, when inaecurate second reaches were promptly followed by screen blanking instead of the desired green screen, a strong burst of impulses was elicited (/•7(/u/-

'

tch

100

i"

lilt 1 r iin 1 " (" 1 'i '

p ' 1 1 II 11 I I I I

• '

."'riMl' II 111 II

1 1

• I 1

'

"

II (

1

1

1

r 1 1 [I n Ml il'.l.

1 11 1

"' w

targ

1 1

go

M l

'„'•',',', r ; "

i i i I' "'';','•,•;'!;'

'in';'. 1

T 1

'

1

11 iii'i 1, 1 , , | | ] 1 1 M 1 1 M rl iri 11 HI |] I ' II II 1 1 1 illlM "111' •Jlri'ii't'i'li'l'lil '•'."'• 111' 1 " l 1

[ II

•

„', ,1 1 iiinmiiiiii •i.'i I ' 1r1

,'.'

,'"1"

' "

111"! I'l ' l l 1 ' r'

'rmnii i m ' i i n'l'ii' 1 ' " I ' '1' ' 1" 1' " ' 1 iir III 1 mi 111 BnU M • 1 iiir 1 ni•! nII 1 1 1 III I 1 1 1 1 M 1 1

1 '1. I ",;

green on

l. (

II M 1 ^

green off reward

B

brank 100

iiMiiiiin I IIIIIMIIIIII mnnii 1

go

blank

1 sec

Figure 4 Example of an area 7a neurone which responded both to the reward itself and to screen blanking which indicated ihal he had missed a reward. The monkey was performing the two-target sequence. {A) The latter part of .successful (rewarded) trials is shown here. A response both to the (second) target presentation (targ) and to the reward is evidenl in the individual trial rasters and the peri-event histogram above. (B) In these trials the first touch (tch) was inaccurate and the screen blanked 450 ms after the presentation of the .second target. {C) For trials in which the second reach was inaccurate the screen blanked instead of turning green; A, B and C are all aligned within the task time frame

significant response to the reward itself, i.e. discharges rates were similar whether a reward was given or not. One of these neurones did, however, respond to target presentation. Discussion Reward-related activity could be confused in some cases with neuronal discharge which was actually linked to a movement, in particular eye-head saccades toward the reward or withdrawal of the arm from the videomonitor. Examples of both of these, but especially the latter, were observed. Accordingly, all neurones with an inerease in discharge at the time of reward delivery were examined for somatosensory input or correlations to arm. eye or head movement. Invariably, regular bursts of activity which appeared to be elicited by the green screen were in fact coupled to the large rightward saeeades whieh elosely followed the flash of green.

212

Ophthal. Physiol. Opt.. 1992, Vol. 12, April

Responses whieh were genuinely linked to the green eue were of the 'anticipation' type. The expectation of events is frequently represented in the firing patterns of posterior parietal neurones^. Thus the discharge of the neurone in Figure 5 could signify the expectation of a reward or. more immediately, the expectation of reeeiving knowledge of his success. The activity was not linked to the motor aspects of touching the screen. For some 'reward' neurones, the response may not have been to the approaching reward itself but to the anticipation of contact with the mouth ^. Since such cells responded whenever any object was brought elose to the mouth as if to touch it, they were identified and excluded from the reward-related category. Motter and Mountcastle'' have found that many area 7a neurones are responsive to moving visual stimuli. The receptive field generally spares the foveai region and shows an opponent vector organization in the peripheral visual field. This would explain our 'reward responses'

Reward respon.ses in parietal cortex: W. A. MacKay et al.

25 1 imp/s

I I HIM II < II IIHIM II in

\\

i\

11 I I II I

n wI I

IJ

I I

I n I I mil

I

II

tl I

I 1 I

I

I

II

I

I

I

I

I I

I I mil

It

I

I I 111 I I I I I 1 1

I 11

i i n I II • I

I

I

I I 111 I

I

I

I M I

I I ^1 imn

I I

I

II

tni

I I'

I

I

III 1

nnii I I i n

Ml

I t

II

I

I I

I

GREEN ON

GO

II

1 II

I

11

II

ni

t I

I

n m f I

t

1 II I n III!

I

I

GREEN OFF

B

25

III III II

II

I

I

nil n I I I II IM

III

I lUMIII

I

II

I il

II I I It

I

II

I

II

III

II

I I

I

I I

I t I

I ' II 111 11

II

1

I

II

I

II in I I I I I

II

I I I II I I II 1 I I mm in I

I I ninii

II

I lit

II

I I

I

t

III

I

I

I 111 • I IK I

n

n i l I I I I

II M l I I 1 I 1 1 II I I I mi I

I

I

\ II

IIII

II

I

II

I I II I

It III II 11(11 I II II 1 1 I I II I N I I I I I I

I

II

tl

I I

I

II

I tl

D i l i

II I

M

I I

II I II II 1 I III I M l II

t t I I I \ I II I I HE I II t n i i t i i i i i n I I I I

III

I tIB

I

GO

n

I

I I M I

I

I

11

I

I

I

I I l U

II I

I

IIII

I II

I

I It I I

I

Ml

I

I

I

I

II 1 I I

NO GREEN

1 sec

Figure 5 A neurone which responded to the expectation of the green background on the videomoniior. \A) Successful trials in which the green field appeared are grouped together here. (B) Unsuccessful trials are grouped here. The discharge rate still increases at about the time of the expected green screen

except that the monkey was usually looking direetly at the reward for mueh of the response. In addition, approaching the monkey repeatedly without a reward gave a gradually habituating response. On the other hand, Rolls and eoworkers^ did not find any influence of motivational set oti area 7 visual or reach-related activity. They reported that aversive, neutral and desirable objects all elicited similar responses. Of course, we eannot conclude that it is the desirability of the reward that generates the response. The significant factor may be

behavioural importance. Through a long period of experience, however, every aspect of the task must have acquired limbic associations^. Area 7a of the parietal lobe is generally associated with the guidance of spatially-directed movcments\ Within area 7a one can find visually-driven neurones which may respond to targets in specific spatialfields**or ceils related to arm'^* or eye movements'V The presence of cells responding to rewards, without regard for spatial attributes seems out of place. It is true, however, that no

Ophthal. Physioi. Opt., 1992, Vol. 12, April 213

Reward responses in parietal eortex: W. A. MacKay et al. motor reaction was required by the monkey at this stage. The reward was delivered direetly to his mouth. Nevertheless, it is ditficult to extract any spatial meaning from the activity of cells with a major reward-related component. Perhaps the activity of such cells serves to draw attention to the occurrence of a behaviourally meaningful event which allows a plan of action to be set in motion. The rapid reaction of a neurone sueh as that in Figure 4 to task cues, reveals an internal set of expectancy, awaiting the resolution of an uncertainty. One may hypothesize that a specific subset of area 7a neurones marks such special events, while possibly another (probably overlapping) subset monitors the spatial properties of the event.

Acknowledgements This work was supported by the Medical Research Council of Canada.

References 1. Slein, J. F. Representation of egocentric space in the posterior parietal cortex. Q. J. Exp. Physiot. 74, 583-606 11989). 2. Cavada. C, and Goldman-Rakic, P. S. Posterior parietal cortex in rhesus monkey: I. Parcellation of areas based on distinctive

214

Ophthal. Physiol. Opt.. 1992, Vol. 12. April

limbic and sen.s()ry corticocorticai connections. J. Comp. Neurol. 287, 393-421 (1989). Mesulam, M.-M,. Van Hoesen, G. W.. Pandya. D. N. and Geschwind, N. Limbic and sensory connections of the inferior parietal lobule (area P G | in the rhesus monkey; a study wilh a new method for horseradish peroxidase histochemistry. Bruin Res.

t36. 393-4!4(1977). MaeKay, W. A.. Mendonga, A. J. and Blum, B. Reaching in the dark: enhanced responses in posterior parietal cortex. NeuroRepori

I, 101-103(1990). MacKay, W. A. and Crammond, D. J. Neuronal correlates in posterior parietal lobe of the expectation of events. Behav. Brain Res. 24, 167-179 (1987). Motter, B. C. and Mountcastle, V. B. The functional properties of the light-sen.sitive neurons of the posterior parietal cortex studied in waking monkeys: foveal sparing and opponent vector organization. ,/. Neuro.sci. I, 3-26 (1981). Rolls. E. T., Perrett. D.. Thorpe, S. J., Puerto. A., Roper-Hall, A. and Maddison, S. Responses of neurones in area 7 of the parietal cortex lo objects of different significance. Brain Res. 169. 194 198 (1979). Bariow, H. B. Unsupervised learning. Neural Computation \, 295-311 (19S9). MacKay. W, A. and Riehk, A. Correlates of preparation of arm reach parameters in parietal area 7a of the cerebral cortex. In Tutorials in .Motor Netiro.'icience (eds. J. Requin and G. E. Stelmach), Kluwer, Dordrecht, pp. 347 356 11991). 10. Blum, B. Manipulation reach and visual reach neurons in the inferior parietal lobule of the rhesus monkey. Behav. Brain Res. t8, 167-173(1985). Robinson, D. L.. Goldberg, M. E. and Stanton. G. B. Parietal association cortex in the primate: sensory mechanisms and behavioral modulations. J. Neurophysiol. 41. 9tO-932 (1978).