Proportion estimation among subsets of features and conjunctions M Bulatova1, I Utochkin2 |
|---|
1Department of Psychology, National Research University 'HSE', Russian Federation |
| Summary statistics is an efficient code that allows representing multiple objects despite severe attentional limitations. Treisman claimed that we can easily represent statistical properties within overlapping feature-marked but not conjunction-marked subsets [Treisman, 2006, Visual Cognition, 14(4-8): 411–443] since conjunctions require focused attention to be bound properly. We tested this prediction thoroughly in three experiments modifying one of Treisman’s experiments. Observers were briefly presented with sets of red, green, or blue Ts, Xs, and Os (either all 3×3 or only arbitrary 2×2 features were used to impose different working memory load) and were to evaluate the proportion of precued or postcued feature (color or shape) or conjunction. Our results showed that both features and conjunctions were estimated equally precisely, which is inconsistent with the results reported by Treisman; though increased number of features led to less accurate evaluation due to certain limitations of visual working memory. Evaluation was more precise in precue condition, except blocking color, shape and conjunction trials, when precue did not affect accuracy regarding features but not conjunctions. This could be because three particular features do not exceed working memory capacities [Halberda et al., 2006, Psychological Science, 17(7): 572-576) and are evaluated easily even when not cued. |
| Up Home |
|---|