The effect of multiple responses and certainty estimates on the integration of visual information

Abstract

was investigated in two experiments that employed a multiple-observation. tilt-discrimination task. The integration model of signal-detection theory generally overestimated the improvement in discrimination resulting from additional observations. Further. probabilistic multiple responses appeared detrimental to discrimination: These results differ from earlier findings using different discrimination tasks. Reasons for the differences were discussed in terms of within-trial redundance and interference causedby complex response modes.

Bayesian aggregation of independent successive visual inputs.

“…The accurate predictions of the entire pattern of results in this study, and in two others (Doherty & Keeley, 1969;Keeley, Doherty, & Bach-man, 1970), is a convincing argument that the normative model is also an appropriate descriptive model, and provides evidence for the assumption of perceptual independence. The fine-grain predictions suggest that the use of a single index, such as d' (Halpern & Ulehla, 1970), provides a much less complete analysis of the perceptual and decision processes underlying an S's performance in multiple-observation studies.…”

The effect of simultaneous and sequential presentation of stimulus dimensions on absolute judgment accuracy

“…For example, Halpern and Ulehla (1970) showed that the tilt discrimination of highly visible lines was improved by sequential stimulus repetition. These findings are consistent with the small increase in the absolute judgment accuracy of brightness, observed in the present study.…”