Combining spatial and temporal expectations to improve visual perception

Abstract

The importance of temporal expectations in modulating perceptual functions is increasingly recognized. However, the means through which temporal expectations can bias perceptual information processing remains ill understood. Recent theories propose that modulatory effects of temporal expectations rely on the co-existence of other biases based on receptive-field properties, such as spatial location. We tested whether perceptual benefits of temporal expectations in a perceptually demanding psychophysical task depended on the presence of spatial expectations. Foveally presented symbolic arrow cues indicated simultaneously where (location) and when (time) target events were more likely to occur. The direction of the arrow indicated target location (80% validity), while its color (pink or blue) indicated the interval (80% validity) for target appearance. Our results confirmed a strong synergistic interaction between temporal and spatial expectations in enhancing visual discrimination. Temporal expectation significantly boosted the effectiveness of spatial expectation in sharpening perception. However, benefits for temporal expectation disappeared when targets occurred at unattended locations. Our findings suggest that anticipated receptive-field properties of targets provide a natural template upon which temporal expectations can operate in order to help prioritize goal-relevant events from early perceptual stages.

Keywords: spatial expectation; synergistic effect; temporal expectation; timing; visual attention.

Figure 1

Schematic illustration of the task structure. Foveally presented cues predicted simultaneously where and when target events were more likely to occur. Cue validity for both spatial and temporal expectation was fixed at 80%. Targets consisted of a horizontally or vertically oriented Gabor patch followed by a backward pattern mask. Targets were presented at a fixed contrast, individually adjusted to equate discrimination performance across individuals. Observers responded to the Gabor orientation with their left of right index finger (counterbalanced across subjects).

Figure 2

Behavioral results. Effects of spatial and temporal expectations on (a) d′, (b) accuracy, and (c) inverse efficiency. Error bars represent SEM.

Figure 3

Effects of temporal expectations on d′ measures at the individual level and permutation tests. Scatter and bagplots showing the main effects of temporal and spatial expectations (Top) and the effect of temporal expectation at the attended and unattended locations (Bottom). Blue indicates the conditions in which there was a significant effect of expectation. The cross at the center of the bagplot represents the center of mass of the bivariate distribution of empirical data, the darker area includes 50% of the data with the largest depth, the lighter polygon contains all other nonoutlier data points, and the Xs represent outliers (Rousseeuw, Ruts, & Tukey, 1999). Outliers were detected using the “Skewness-Adjusted Outlyingness” method using LIBRA toolbox for MATLAB (Verboven & Hubert, ; for more details on this method see Hubert & Van der Veeken, 2008). Histograms showing the distributions generated from 10,000 random permutations from data sets of 20 participants (see Methods for details). Original t test values (Attended vs. Unattended time at Unattended (left) and Attended (right) Locations) are indicated by the black vertical lines.