Peter Scarfe¹, Anna Horwood², Patricia Riddell³; ¹School of Psychology and Clinical Language Sciences, ²University of Reading, ³Reading, UK

Contemporary models of sensory cue integration propose that by combining sensory cues we gain a more precise sensory estimate compared to that available from the cues in isolation. This is achieved by weighting cues in proportion to their relative reliability, and their reliability relative to a priori information about the perceptual property being estimated. Cue weighting is typically measured behaviourally, e.g. via a two alternative forced choice experiment. However, theoretically, one could determine cue weighting using data from any level of the perceptual hierarchy, e.g. from physiological response through to perceptual decision. Here, we derive equations allowing one to test cue integration using cue weights inferred from the gain function relating physiological responses to a varying driving stimulus. We apply this model to an extensive preexisting data set of vergence and accommodative responses (measured with a Remote Haploscopic Photorefractor) to distance targets (0.5, 1, 2 and 3 dioptres) varied to isolate individual cues of disparity, blur and angular subtense (Horwood and Riddell, 2008). We find that combined cues physiological gain functions are well predicted by that of individual cues. This being clearer for vergence compared to accommodative responses. Our results offer a novel way in which to test models of sensory cue integration, which complements current behavioural methods. We highlight the assumptions and limitations of this approach, as well as its ability to disambiguate between competing models of cue integration.