Fig. 1.

Constructing an orthonormal basis for a delayed-match-to-sample (DMS) task. A: each trial of the DMS task began with the presentation of a cue indicating the target for that trial, followed by the presentation of 0–3 distractors and then the target match. Images were presented for 400 ms, followed by a 400-ms blank. Monkeys were required to maintain fixation throughout the distractors and saccade to a response dot after the target match appeared (within 800 ms) to receive a reward. B: the experimental design included 4 images, each presented as a visual stimulus (“looking at”) in the context of every other image as a target (“looking for”), thus defining a 4 × 4 matrix. In this matrix, target matches fall along the diagonal and distractors fall off the diagonal. C: the matrices produced by the first stage of the orthonormal basis design process (see text). D: the matrices produced by applying the Gram-Schmidt process to the matrices described in C. E: an experimental design in which the “target match” and “visual” conditions cannot be orthogonalized (see text).