Abstract
When viewing an elephant above a bus, pFs represents both objects simultaneously and is far more sensitive (evidenced by greater release of BOLD adaptation) to a change in the relative positions of the objects (bus now depicted above the elephant) than to a scene translation with equal object displacements. Global features, eye movements, and task difficulty were ruled out as possible explanations of this effect (Hayworth et al., 2008). If the relation changes could be tracked, would they still produce a release from adaptation?
Subjects viewed a 200ms S1 consisting of two separated objects each enclosed by a box outline, followed by a 300ms dynamic rearrangement in which the objects disappeared but the empty boxes remained and moved smoothly to new screen positions, followed by a 200ms S2 where objects reappeared within the stationary boxes. Objects always appeared at constant eccentricity. Box movement and the congruency of their contents defined four trial types: SameDirCongruent: boxes moved in same direction, S2 objects reappeared in the same boxes they did in S1, (e.g., elephant above bus -[[gt]] elephant above bus); SameDirIncongruent: boxes moved in the same direction, but objects swapped boxes between S1 and S2 (e.g. elephant above bus -[[gt]] bus above elephant); OppDirCongruent: boxes moved in opposite directions, with the S2 objects remaining in the same boxes as S1 (e.g. elephant above bus -[[gt]] elephant side-of bus), OppDirIncongruent: boxes moved in opposite directions, objects swapped boxes (e.g. elephant above bus -[[gt]] elephant side-of bus). In both pFs and IPS only incongruence resulted in a sizable release from adaptation over SameDirCongruent OppDirCongruent gave only minimal release even though objects changed relative positions (e.g. above to side-of) between S1 and S2. These results are consistent with Kahneman and Treisman's (1984) Object File model but pose a challenge to simple stimulus-driven interpretations of adaptation.
NSF BCS 04-20794, 05-31177, 06-17699