Abstract
Recent studies of object color perception suggest that semantic information affects color appearance (Hurlbert & Ling, VSS2005; Gegenfurtner & Walter, ECVP2004). We examine whether semantic primes presented 300 msec before a test square alter the square's color appearance. We separated prime and test square in time to insure that any effect would be semantic and not due to the shape or spatial frequency content of the primes. Preliminary: Each observer first completed tasks that allowed us to construct an equiluminant red-green scale (passing through neutral) specific to the observer. Task: On each trial, observers saw a briefly-presented prime followed after 300 msec by a briefly-presented test square whose chromaticity fell on the red-green scale. The task was to judge whether the test square was “reddish” or “greenish.” Chromaticity was adjusted in a staircase procedure along the scale to converge on the point of subjective achromaticity (PSA). There were four prime conditions, each with separate staircases: red- and green-associated words and red- and green-associated images. Psychometric functions were fitted to estimate the PSA for each condition. Five naïve observers participated. Results: For four observers, PSAs for the red and green word-primed conditions did not differ significantly. For three observers, PSAs for the red and green picture-primed conditions did not differ. The remaining significant differences were opposite in direction to the predictions of semantic priming (i.e. PSAs for red primed conditions were greenish). Conclusion: The only effect of achromatic semantic primes was to push surface color judgments away from the primed color.