Neuroscience: Formation of a Percept in the Rat Cortex

Dissecting the Roles of Supervised and Unsupervised Learning in Perceptual Discrimination Judgments

Our ability to compare sensory stimuli is a fundamental cognitive function, which is known to be affected by two biases: choice bias, which reflects a preference for a given response, and contraction bias, which reflects a tendency to perceive stimuli as similar to previous ones. To test whether both reflect supervised processes, we designed feedback protocols aimed to modify them and tested them in human participants. Choice bias was readily modifiable. However, contraction bias was not. To compare these results to those predicted from an optimal supervised process, we studied a noise-matched optimal linear discriminator (Perceptron). In this model, both biases were substantially modified, indicating that the “resilience” of contraction bias to feedback does not maximize performance. These results suggest that perceptual discrimination is a hierarchical, two-stage process. In the first, stimulus statistics are learned and integrated with representations in an unsupervised process that is impenetrable to external feedback. In the second, a binary judgment, learned in a supervised way, is applied to the combined percept.

SIGNIFICANCE STATEMENT The seemingly effortless process of inferring physical reality from the sensory input is highly influenced by previous knowledge, leading to perceptual biases. Two common ones are contraction bias (the tendency to perceive stimuli as similar to previous ones) and choice bias (the tendency to prefer a specific response). Combining human psychophysical experiments with computational modeling we show that they reflect two different learning processes. Contraction bias reflects unsupervised learning of stimuli statistics, whereas choice bias results from supervised or reinforcement learning. This dissociation reveals a hierarchical, two-stage process. The first, where stimuli statistics are learned and integrated with representations, is unsupervised. The second, where a binary judgment is applied to the combined percept, is learned in a supervised way.

A sensory integration account for time perception

The connection between stimulus perception and time perception remains unknown. The present study combines human and rat psychophysics with sensory cortical neuronal firing to construct a computational model for the percept of elapsed time embedded within sense of touch. When subjects judged the duration of a vibration applied to the fingertip (human) or whiskers (rat), increasing stimulus intensity led to increasing perceived duration. Symmetrically, increasing vibration duration led to increasing perceived intensity. We modeled real spike trains recorded from vibrissal somatosensory cortex as input to dual leaky integrators-an intensity integrator with short time constant and a duration integrator with long time constant-generating neurometric functions that replicated the actual psychophysical functions of rats. Returning to human psychophysics, we then confirmed specific predictions of the dual leaky integrator model. This study offers a framework, based on sensory coding and subsequent accumulation of sensory drive, to account for how a feeling of the passage of time accompanies the tactile sensory experience.

The limited capacity of visual temporal integration in cats

It has been long known that prolonging stimulus duration may increase the perceived brightness of a visual stimulus. The interaction between intensity and duration generally follows a rule, such as that described in Bloch's law. This visual temporal integration relationship has been identified in human subjects and in non-human primates. However, although auditory temporal integration has been extensively studied in the cat, visual temporal integration has not. Therefore, the goal of this study was to examine visual temporal integration in the cat. We trained five cats to respond when a brief luminance change was detected in a fixation dot. After training, we measured the success rate of detecting the luminance change with varying durations at threshold, subthreshold, and suprathreshold luminance levels. Psychometric functions showed that prolonging stimulus duration improved task performance, more noticeably for stimuli below 100 ms than beyond. Most psychometric functions were better fit to an exponential model than to a linear model. The gradually saturated performance observed here, as in previous studies, can be explained by the “leaky integrator” hypothesis, that is, temporal integration is only valid below a critical duration. Overall, we developed a task whereby visual temporal integration was successfully demonstrated in the cat. The effect of stimulus duration on detection success rate displayed a pattern generally consistent with previous human and non-human primate findings on visual temporal integration.