Purchase this article with an account.
E. Greene; Evidence That Contour Attributes Are Not Essential for Recognition of Objects. Invest. Ophthalmol. Vis. Sci. 2007;48(13):5891.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
This hypothesis was tested using a minimal transient discrete cue (MTDC) protocol, in which spaced dots that mark the boundary of an object are used to elicit recognition of the object.
For each object, a sample was selected from the total complement of boundary dots, designated as the display set, the size of this set being larger or smaller to provide for approximately equal recognition across the inventory of objects. The display set was further broken into subsets, each with 4 dots, requiring either that the dots be spatially adjacent (forming a short string) or picked at random from among the border positions of the display set. Each dot was shown successively for 0.1 ms, with 0 ms of separation between members of the subset. The location of successive subsets was chosen at random. Time intervals were added between successive subsets, ranging from 0-6 ms.
These task conditions produced declines in recognition that were equivalent for adjacent vs. random subsets. The adjacent subsets would be expected to activate orientation selective cells, whereas random subset would not.
These findings argue against the concept that contour attributes provide essential cues for recognition. An alternative that calls for registering the pattern of boundary positions will be outlined.
This PDF is available to Subscribers Only