Decoding of coherent but not incoherent motion signals in early dorsal visual cortex

Abstract

When several scattered grating elements are arranged in such a way that their directions of motion are consistent with a common path, observers perceive them as belonging to a globally coherent moving object. Here we investigated how this coherence changes the representation of motion signals in human visual cortex using functional magnetic resonance imaging (fMRI) and multivariate voxel pattern decoding, which have the potential to reveal how well a stimulus is encoded in different contexts. Only during globally coherent motion was it possible to reliably distinguish fMRI signals evoked by different directions of motion in early visual cortex. This effect was specific to the retinotopic representation of the visual field quadrant in V1 traversed by the coherent element path and could not simply be attributed to a general increase in signal strength. Decoding was more reliable for cortical areas corresponding to the lower visual field. Because some previous studies observed poorer speed discrimination when motion was grouped, we also conducted behavioural experiments to investigate this with our stimuli, but did not reveal a consistent relationship between coherence and perceived speed. Taken together, these data show that neuronal populations in early visual cortex represent information that could be used for interpreting motion signals as unified objects.

Fig. S1

Flat maps from a representative participant. A. The response to all stimuli in the main experiment relative to blocks of fixation is plotted at a relaxed statistical threshold (p < 0.05, uncorrected). B–C. Voxels in V1d (representing the curve quadrant) showing a moderately significant bias for direction of motion (p < 0.1, uncorrected) in the coherent context (B) and the incoherent context (C). The dotted black lines indicate the boundaries of visual areas. The yellow dotted line denotes the location of the ROI defined by the ‘element localizer’. The mean decoding accuracy (± standard error of the mean) for direction of motion in V1d across all instances of cross-validation is denoted above each map.

Fig. S2

Psychometric curves in the behavioural experiment for each of the four participants (in rows). We plotted the proportion of trials in which the test stimulus was perceived as faster than the reference as a function of the veridical ratio between test and reference speed (in logarithmic units). Thus a rightward shift of the curve would indicate that participants saw the stimulus as moving more slowly than the reference stimulus. Data are shown either for pooling data from the two possible positions of the curve quadrant (left), or separately for when the curve quadrant was the lower-right (middle) or upper-left quadrant (right). Symbols indicate raw data; the solid lines are the best fitting models. Blue: coherent context. Red: incoherent context. Green: control (single element in both reference as well as test stimulus).

Fig. 1

Static illustrations of the stimuli used. Five Gabors were placed on an imaginary curved path. Elements were either oriented in a coherent (A–B) or incoherent (C–D) relationship with the global context of the path. A sixth Gabor (the distractor element) was located in the diametrically opposite location to the middle element of the curve (the curve element). Within separate scanning sessions the individual elements making up the curve either traversed the lower-right (A, C) or the upper-left (B, D) visual field. Movies of the actual stimuli used in the experiment are included in the supplementary materials.

Fig. 2

Discriminating the coherent versus incoherent contexts using voxel patterns from visual areas. Decoding accuracy averaged across participants is depicted for each ROI separately for when the curve traversed the lower-right visual field (A) or the upper-left visual field (B). Error bars denote ± 1 standard error of the mean. Asterisks indicate accuracy was significantly higher than chance (one-tailed t-test, p < 0.05). The arrows in the stimulus schematics indicate the directions of motion. Black arrows: coherent context. Grey arrows: incoherent context.

Fig. 3

Decoding the direction of motion from voxel patterns in human visual cortex. Accuracy averaged across participants is plotted for retinotopic areas V1–V3 (error bars denote ± 1 standard error of the mean). The plots are superimposed on a stimulus schematic to indicate the ROI (i.e. the curve quadrant traversed by the global context or the distractor quadrant in the opposite hemifield). In separate experiments, the curve quadrant was either the lower-right (A) or the upper-left (B) visual field. Black arrows/bars: coherent context. Grey arrows/bars: incoherent context.

Fig. 4

Mean BOLD signals evoked in early visual cortex by the globally coherent and incoherent contexts. Mean parameter estimates from the GLM analysis of all visually responsive voxels (see Materials and methods; averaged across participants) are plotted for regions of interest in early visual cortex. Panels are superimposed on a stimulus schematic indicating the ROI (curve quadrant and distractor quadrant). Only data from experiments when the curve quadrant was the lower-right visual field are shown. Black bars: coherent context. Grey bars: incoherent context. Error bars denote ± 1 standard error of the mean.

Fig. 5

Behavioural experiment. The points of subjective equality for judging whether the Gabors moved faster than a reference element are plotted for each experimental condition and each of the four participants. Values indicate the ratio between the test and reference speeds (in logarithmic units). Thus, positive numbers indicate that participants saw the test interval as moving slower than the reference interval. Different symbols denote the data from individual participants. All data here are pooled across the two possible locations of the curve quadrant (lower-right and upper-left). Psychometric curves for individual observers and global context locations are shown in Fig. S2.