Predicting human perceptual decisions by decoding neuronal information profiles

Abstract

Perception relies on the response of populations of neurons in sensory cortex. How the response profile of a neuronal population gives rise to perception and perceptual discrimination has been conceptualized in various ways. Here we suggest that neuronal population responses represent information about our environment explicitly as Fisher information (FI), which is a local measure of the variance estimate of the sensory input. We show how this sensory information can be read out and combined to infer from the available information profile which stimulus value is perceived during a fine discrimination task. In particular, we propose that the perceived stimulus corresponds to the stimulus value that leads to the same information for each of the alternative directions, and compare the model prediction to standard models considered in the literature (population vector, maximum likelihood, maximum-a-posteriori Bayesian inference). The models are applied to human performance in a motion discrimination task that induces perceptual misjudgements of a target direction of motion by task irrelevant motion in the spatial surround of the target stimulus (motion repulsion). By using the neurophysiological insight that surround motion suppresses neuronal responses to the target motion in the center, all models predicted the pattern of perceptual misjudgements. The variation of discrimination thresholds (error on the perceived value) was also explained through the changes of the total FI content with varying surround motion directions. The proposed FI decoding scheme incorporates recent neurophysiological evidence from macaque visual cortex showing that perceptual decisions do not rely on the most active neurons, but rather on the most informative neuronal responses. We statistically compare the prediction capability of the FI decoding approach and the standard decoding models. Notably, all models reproduced the variation of the perceived stimulus values for different surrounds, but with different neuronal tuning characteristics underlying perception. Compared to the FI approach the prediction power of the standard models was based on neurons with far wider tuning width and stronger surround suppression. Our study demonstrates that perceptual misjudgements can be based on neuronal populations encoding explicitly the available sensory information, and provides testable neurophysiological predictions on neuronal tuning characteristics underlying human perceptual decisions.

Fig. 1

a Illustration of the visual motion stimulus used for the center-surround paradigm in the experiment and model. b Theoretical psychometric function of a subject for discriminating the direction of motion of the center target from the vertical (upward) reference direction. It represents the proportion of “rightward” answers as a function of the center motion direction, and helps to visualize the perceived vertical reference direction (midpoint) as well as the discrimination threshold for reliably (above p=0.84) seeing a deviation from the perceived reference. Theoretical models need to predict the perceived value of the stimulus. The discrimination threshold is the error on the perceived value

Fig. 2

Illustration of population activity and Fisher information (FI) within a theoretical neuronal population and its hypothetical modulation by surround motion. a Schematic of a population of neurons sensitive to all directions of motion and arranged in a hypercolumn of motion sensitive neurons in macaque cortical area MT/V5. b The corresponding theoretical population response function (thick line) based on neurons with all possible preferred directions of motion in response to a zero degree motion direction. Error bars denote the Poisson variability of the firing rate of the neurons. The thin line represents the FI information profile of the neuronal population. c Modulation of neuronal tuning curve amplitudes induced by a second motion stimulus in the surround moving +40° away from zero degree. The suppressive influence on the amplitude is shown in thick grey (Eq. 2). d The corresponding population activity and FI profile modulated by the presence of the surround (thick and thin line, respectively). The FI estimate is shown as vertical white line and the Standard models estimate as vertical black line (A _i0 = 0.8, σ = 30). Note that both models provide different estimates (the x-position of the vertical lines) given identical population characteristics

Fig. 3

a Illustration of the prediction of the perceived reference direction (0°) for the FI (thick black line) and the Standard models (thin grey line) as a function of the distracting surround motion direction (different y-scales for the two curves). b Elevation of the discrimination thresholds as a function of surround motion direction (ratio between discrimination threshold for the center-with-surround stimulus and the center-without-surround stimulus). An elevation of 1.0 indicates that the discr. threshold center-with-surround is not different from center-without-surround’s discr. thresholdwhen adding a surround with the corresponding motion direction (x-axis)

Fig. 4

a Example of a staircase run which started by presenting a target direction of 21° and “walks” over up to the 75% convergence point where it has a typical random walk (condition with surround motion of −70°). b The corresponding pooled responses (dots), i.e. proportions of “leftward” answers as a function of target direction of motion together with the fitted logistic function (solid curve; a = −5.93, b = −0.44). The number of trials at each visited target motion direction are shown above/below each datum. This example shows the repulsion effect on the psychometric function, with the midpoint shifted closer to the surround, such that the subject responded to the physical vertical direction (0°) to be further away from the surround. The psychometric function allowed to extract the discrimination threshold as illustrated in Fig. 1b. c Histogram of all target motion directions presented across subjects. The distribution peaks around 0° (mean: −0.073; SE: 0.054; n = 21, 600), demonstrating that globally the staircase method presented a mean upward target. Therefore the mathematical condition for applying the MAP estimate is met (see text for details)

Fig. 5

Psychophysical results and model fit. a Average psychophysical repulsion curve (dots) obtained by computing the between-subjects mean perceived vertical upward reference direction of the target motion as a function of the direction of the surround motion. The model fits are shown as solid line: FI model in black, standard models in grey. Light grey data points represent deviations from the predicted motion repulsion (see Sect. 4). b Average normalized discrimination thresholds of the target motion (dots) and model fit (line) as a function of the surround motion direction. Error bars denote SE (n = 20)

Fig. 6

Examples of typical motion repulsion effects in two subjects. Shown is the physical motion direction perceived to be the vertical upward direction (dots), as a function of the surround motion direction. Each graph also presents the fitted FI model together with the resulting parameters (the fits did not include the four data points represented with squares, see main text)

Fig. 7

Results of fitting the data of Kim and Wilson (1997) (extracted from their Fig. 3) with the FI and standard models (black and grey solid line, respectively). Error bars are SE (n = 4)