Visual short-term memory operates more efficiently on boundary features than on surface features.

摘要: A change detection task was used to estimate the visual short-term memory storage capacity for either orientation or size of objects. On each trial, several objects were briefly presented, followed by a blank interval and then second display that identical first had single object different (the changed size, in separate experiments). The indicate whether two displays same different, number remembered estimated from percent correct on this task. Storage feature nearly twice as large when defined boundary, rather than surface texture object. This dramatic difference suggests particular (e.g., right tilted small) is not stored with an invariant abstract code. Instead, there appear be codes boundary features objects, operates more efficiently it features.