Separate signals for target selection and movement specification in the superior colliculus

Science

Volumn 284, Issue 5417, 1999, Pages 1158-1161

  Horwitz, Gregory D ^a   Newsome, William T ^a  

^a STANFORD UNIVERSITY SCHOOL OF MEDICINE   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

ANIMAL CELL; ARTICLE; EVOKED VISUAL RESPONSE; NERVE STIMULATION; NONHUMAN; PRIORITY JOURNAL; SACCADIC EYE MOVEMENT; SIGNAL DETECTION; SUPERIOR COLLICULUS; VISUAL DISPLAY UNIT; VISUAL STIMULATION;

ANIMALS; MACACA MULATTA; MOTION PERCEPTION; NEURONS; PHOTIC STIMULATION; SACCADES; SUPERIOR COLLICULUS; VISUAL PERCEPTION;

ANIMALIA;

EID: 0033553667     PISSN: 00368075     EISSN: None     Source Type: Journal    
DOI: 10.1126/science.284.5417.1158     Document Type: Article

References (29)

1. P. H. Schiller and F. Koerner, J. Neurophysiol. 34, 915 (1971).
2. R. H. Wurtz and M. E. Goldberg, Science 171, 82 (1971).
3. _, J. Neurophysiol. 35, 575 (1972).
4. D. L. Sparks, Brain Res. 156, 1 (1978).
5. L. E. Mays and D. L. Sparks, J. Neurophysiol. 43, 207 (1980).
6. P. W. Glimcher and D. L. Sparks, Nature 355, 542 (1992).
7. D. P. Munoz and R. H. Wurtz, J. Neurophysiol. 73, 2313 (1995).
8. Three monkeys were trained to discriminate opposite directions of motion in a stochastic visual display centered on the fovea. The stimulus was a circular field of randomly positioned dots, a proportion of which translated coherently in a specified direction whereas the remainder were replotted at random locations on successive video frames. On each trial, the monkey reported the direction of motion of the coherent dots by making a saccade to one of two targets presented at either end of the motion axis. The difficulty of the task was pseudo-randomly varied from trial to trial by changing the proportion of coherently moving dots. The monkey received liquid rewards for choosing the target in the direction of coherent motion.
9. While the monkeys performed the task, we qualitatively assessed prelude activity in 704 intermediate and deep-layer SC neurons. Roughly 25% of the cells exhibited target-specific prelude activity several seconds before saccade execution. We obtained quantitative data sets from 127 neurons. For each of these cells, we counted spikes during the presentation of the motion stimulus and the first second of the delay period. The counts were normalized within each stimulus coherence and compiled into two distributions according to the monkey's choice in the discrimination task (T1 or T2). For each cell, the difference between these two distributions was evaluated by a Mann-Whitney U-test with a criterion level of P < 0.01. By this criterion, 103 of 127 cells exhibited predictive activity. Seven of these cells were eliminated from further analysis because they were significantly more active preceding saccades to T2 than to T1. Thus, our final database consisted of 96 choice-predicting SC neurons.
10. K. H. Britten, M. N. Shadlen, W. T. Newsome, J. A. Movshon, J. Neurosci. 12, 4745 (1992).
11. J. D. Roitman and M. N. Shadlen, Soc. Neurosci. Abstr. 24, 262 (1998).
12. For each cell isolated, we presented 6 to 15 repetitions of a 51.2% coherence stimulus of 1- to 2-s duration, moving either toward or away from the movement field.
13. In saccade trials, a single target was illuminated contralateral to the cell's movement field 300 ms after the monkey fixated. This event indicated to the monkey that a saccade to the target (within 500 ms of fixation point disappearance) would be required in order to obtain a reward. Saccade and passive fixation trials were randomly interleaved.
14. 2)]
15. N. P. Bichot, J. D. Schall, K. G. Thompson, Nature 381, 697 (1996).
16. This monkey was trained to perform a direction discrimination task in which the mapping between the direction of visual motion and the operant saccade vector varied substantially from trial to trial. Direction selectivity was measured during a passive fixation task, as it was for the other monkeys in this study. Only 5 of 35 intermediate-layer neurons exhibited significant direction selectivity (Mann-Whitney U-test: P < 0.05). Data from this animal will be described fully elsewhere.
17. We performed this analysis separately for each motion strength using only correctly answered trials. Time was quantized in nonoverlapping 100-ms bins spanning the trial. We calculated normalized activity for each cell by counting the number of spikes in each time bin and dividing by the maximum mean spike count (mean across trials, maximum across motion strengths and time bins). Normalized spike counts were first pooled across cells and then segregated with respect to the monkey's psychophysical decision. For each time bin, we calculated an ROC (receiver operating characteristic) curve from the two distributions of spike counts and integrated its area. This metric can be interpreted as the probability with which an ideal observer could predict the monkey's choice on the basis of the normalized spike count in that time bin.
18. D. M. Green and J. A. Swets, Signal Detection Theory and Psychophysics (Wiley, New York, 1966).
19. M. N. Shadlen and J. N. Kim, Nature Neurosci. 2, 176 (1998).
20. M. Shadlen and W. Newsome, Proc. Natl. Acad. Sci. U.S.A. 93, 628 (1996).
21. For each of the five 100-ms-long epochs preceding the onset of the visual stimulus, we assigned spike counts from T1 choice trials and T2 choice trials randomly to two distributions. We then calculated the ROC curve for the two distributions and integrated its area. This procedure was repeated 5000 times to estimate the distribution of ROC areas that would be expected by chance. The cited P value indicates the proportion of the 5000 iterations in which the ROC value computed from the randomized data equaled or exceeded the ROC value computed from the experimental (nonrandomized) data.
22. We suggest that the monkey enters some trials favoring one target over the other, and that this bias is reflected in small firing rate differences before onset of the visual stimulus. When the motion coherence is high, the direction of stimulus motion largely determines the monkey's choices, overriding the weak bias that exists entering the trial. Because the direction of stimulus motion is chosen randomly, trials in which the monkey is biased toward T1 or T2 have an equal probability of resulting in a T1 or T2 decision, and little or no differential activity is apparent when the trials are subsequently sorted and analyzed according to decision outcome. At 0% coherence, however, the monkey tends to choose in the direction of the bias because there is no directional signal to override the bias. Thus, the small differences in neural activity associated with the bias (before stimulus onset) become associated with the decision and are apparent when the trials are sorted and analyzed according to decision outcome.
23. All coherences were averaged together for these analyses. Latency was defined as the time from stimulus onset until predictive activity exceeded the baseline level by three standard deviations. Time was quantized in 10-ms bins for this analysis because the difference in latency between the two groups was only 30 ms. For the time course analysis we calculated the time to reach half of the maximum predictive activity obtained during the stimulus presentation. This time differed between the two groups by 300 ms. Distributions of both statistics under the null hypothesis were generated by randomly reassigning the cells to two groups 2000 times, calculating the value of the statistic for both groups, and recording the difference. The cited P value is the proportion of differences greater than or equal to the actual difference obtained from the nonrandomized data.
24. M. A. Basso and R. H. Wurtz, J. Neurosci. 18, 7519 (1998).
25. J. W. Gnadt and R. A. Andersen, Exp. Brain Res. 70, 216 (1988).
26. J. Schall and D. Hanes, Nature 366, 467 (1993).
27. P. Mazzoni, R. Bracewell, S. Barash, R. Andersen, J. Neurophysiol. 76, 1439 (1996).
28. D. A. Robinson, IEEE Trans. Biomed. Eng. 10, 137 (1963).
29. We thank C, Barberini, G. DeAngelis, J. Liu, J. Nichols, and M. Shadlen for critical feedback and comments on the manuscript. All experimental procedures and care of the animals were carried out in compliance with guidelines established by NIH and approved by the Institutional Animal Care and Use Committee of Stanford University. Supported by the National Eye Institute (grant 05603) and by the Human Frontiers Science Research Program. G.D.H. was supported by a predoctoral fellowship from the Office of Naval Research and by training grant 5T32NH17047-17 from NIH. W.T.N. is an Investigator with the Howard Hughes Medical Institute.