Decision-related activity in sensory neurons reflects more than a neuron's causal effect

Abstract

During perceptual decisions, the activity of sensory neurons correlates with a subject's percept, even when the physical stimulus is identical. The origin of this correlation is unknown. Current theory proposes a causal effect of noise in sensory neurons on perceptual decisions, but the correlation could result from different brain states associated with the perceptual choice (a top-down explanation). These two schemes have very different implications for the role of sensory neurons in forming decisions. Here we use white-noise analysis to measure tuning functions of V2 neurons associated with choice and simultaneously measure how the variation in the stimulus affects the subjects' (two macaques) perceptual decisions. In causal models, stronger effects of the stimulus upon decisions, mediated by sensory neurons, are associated with stronger choice-related activity. However, we find that over the time course of the trial these measures change in different directions-at odds with causal models. An analysis of the effect of reward size also supports this conclusion. Finally, we find that choice is associated with changes in neuronal gain that are incompatible with causal models. All three results are readily explained if choice is associated with changes in neuronal gain caused by top-down phenomena that closely resemble attention. We conclude that top-down processes contribute to choice-related activity. Thus, even forming simple sensory decisions involves complex interactions between cognitive processes and sensory neurons.

FIGURE 1

Methods. a Schematic of the sequence of events during one trial. b Example time series of the stimulus. c Probability mass distributions of the stimuli for one experiment (probability as a function of disparity), with signal disparities: −0.3° and 0.15°. Each panel depicts one signal condition (negative percentages indicate near signal disparities). d The monkey’s performance for this experiment(in percent near choices as a function of % added signal).

FIGURE 2

Psychophysical kernel and choice-related signal have different time-courses. a Psychophysical kernel (averaged over 76 experiments; n=17200 trials; two monkeys) as a function of disparity and time. Color represents amplitude (in occurrences/frame). b Normalized amplitude of the psychophysical kernels decreases over time. c Averaged choice-related signal over time. b,c Shaded gray areas: ± 1 standard error. d The correlation coefficient (R) over time between CP (for individual neurons) and the amplitude of the mean psychophysical kernel against a neuron’s mean CP. Color-code: temporal integration time (supplementary methods); bold edges: significant R (p<0.05, by resampling); circles, squares data from monkey 1 and 2, respectively.

FIGURE 3

Choice-dependent gain change. a Sub-space maps for preferred (red), null (blue) choices superimposed (neuron d1456). Dashed lines: 0° disparity, 0 spikes/frame. b Null-choice responses plotted against preferred-choice responses, yielding relative gain (slope, 1.84), additive change (y-offset, −0.032 spikes/frame). Dashed lines: unity, 0 spikes/frame. c,d Same format as a,b for neuron d1394 whose slope (1.18), y-offset (0.005 spikes/frame) resemble the population-mean. e Slope and choice-probability are correlated. Filled, open symbols: cells with, without significant choice-probability. Circles, squares: data for monkey 1, 2. Dashed lines: 0.5 choice-probability, relative gain of 1. f No correlation between y-offset and choice-probability (symbols as e). Dashed lines: 0.5 choice-probability, 0 spikes/frame. Solid lines in e,f: median standard error for slope, y-offset.

FIGURE 4

Reward size affects behavior and choice-related signal. a Psychophysical kernel as a function of disparity and available reward (in occurrences/1000ms; n=6886 trials for large reward, red line; n=10314 trials for small reward, blue line; averaged over the first and second 1000ms of each trial in the left and right panel, respectively.). Improved performance is mainly caused by a larger psychophysical kernel in the first (kernel difference p<0.001, by resampling), not second half of the trials (difference n.s.) b Choice-probability computed for the first half of the trials was larger when a smaller reward was available (p<0.006, n=76). Dashed line: unity.