Figure 1.

Recruitment of feature-specific visual representations during working memory maintenance. (A) Data illustrating that the orientation of a to-be-remembered line grating can be decoded from multi-voxel activity patterns measured from visual areas V1-V4 throughout the entire duration of the working memory delay-period. The diagram below the graph depicts the presentation times of the two sample gratings, the ensuing cue (indicating which of the two gratings should be maintained), and the final test grating (upon which participants make their memory-based judgment). The classifier’s ability to decode which orientation was maintained in working memory (green circles) was statistically indistinguishable from its ability to decoded the orientation of a perceived grating that was presented throughout the entire trial (red triangles). Adapted from Harrison & Tong (2009). (B) Delay-period activity patterns from V1 contain sufficient information to decode which of two orientations a participant was maintaining in working memory (on trials in which participants were cued to remember orientation), as well as to decode which of two colors participants were maintaining (on trials in which participants were cued to remember color). In both cases, decoding performance was at chance for the irrelevant stimulus dimension. Adapted from Serences et al. (2009).