Tachistoscopic simulations of eye fixations on pictures

In three picture recognition experiments, complex pictures were presented during a study phase, each presentation consisting of n sequential masked presentations and each presentation lasting d msec. This procedure was designed to mimic a series of eye fixations over a picture, but with number and duration of fixations under experimental control. With n held constant, subsequent recognition memory performance increased with increasing d up to 400 msec. With d held constant, performance increased with increasing n only to the degree that an additional presentation of a picture was used to fixate a novel portion of the picture. These results, and those of Loftus's 1972 experiment, suggest a model of picture encoding that incorporates the following propositions: (a) A normal fixation on a picture is designed to encode some feature of the picture. (b) The duration of fixation is determined by the amount of time required to carry out the intended feature encoding. (c) The more features are encoded from a picture, the better the recognition memory will be form the picture. Additionally, the results of the present experiments imply that the events that constitute encoding within a fixation proceed in a fixed, relatively immutable order.