Reviewing Evidence for Perception-Action Dissociations

Abstract

The Perception-Action Model, proposed by Goodale and Milner (1992), is an influential model about the functional organisation of the human visual system. The core tenet of this model is that visual information is processed in anatomically and functionally separate pathways depending on the output requirements. Object identification and recognition, or broadly visual perception, presumably relies on representations based on allocentric coordinates, relative metrics and holistic processing of the ventral cortical stream. Conversely, visuomotor actions are assumed to rely on egocentric coordinates, absolute metrics and analytic processing of the dorsal cortical stream. Neuropsychological patient studies and studies on the behaviour of healthy participants were initially cited as converging evidence for this model. Subsequent research has shown that inferences from patient data were not as clear-cut as originally reported. Therefore, recent research has focused on demonstrating perception-action dissociations, or differences in performance in perception and action, in the behaviour of neurologically intact individuals.

This dissertation contributes to the work on perception-action dissociations by evaluating the evidence from three different behavioural paradigms. We first examined the claim that grasping violates a fundamental psychophysical principle, Weber’s law (Ganel et al., 2008), demonstrated in several sensory domains including visual perception. We demonstrated that a mathematical flaw in the calculation of the just-noticeable-differences (JNDs), that are used to assess Weber’s law, resulted in an apparent violation of Weber’s law in grasping. We proposed an improved method to estimate JNDs in grasping. Applying this method to our own data and reanalysing the data from three representative studies, we showed that grasping does follow Weber’s law, as does perception. We therefore found no evidence for a perception-action dissociation in Weber’s law.

Next, we looked at the prominent claim that grasping evades Garner interference while speeded-classification and manual size estimation (perception) suffer this interference (Ganel & Goodale, 2003). We reviewed the literature on this and found an empirical inconsistency: there were only two studies that actually demonstrated Garner interference in manual estimation. We thus performed four replications and a quantitative review of fifteen studies on this topic. Our results suggest that Garner interference in grasping and manual estimation are quite similar, although smaller than in speeded-classification. Again, we found no evidence for a perception-action dissociation in Garner interference.

Finally, we investigated the report that grasping is more accurate at discriminating object sizes than perception (Ganel et al., 2012). This conclusion was based on a significant difference between participants’ grip apertures to a small and large object, but a low perceptual judgement accuracy – an unequal comparison of different measures in perception and action. But, when calculating the corresponding classification accuracies in two experiments, we found that grasping has a close-to-chance accuracy and that perceptual judgement accuracies are considerably higher. Manual size estimation accuracy was also close-to-chance and similar to grasping. A meta-analysis of the different studies revealed a similar pattern. Here too, we found no evidence for a perception-action dissociation in size resolution.

Based on our comprehensive investigations including replications, meta-analyses and improved methods, we conclude that neither of the three paradigms under scrutiny here provide evidence for perception-action dissociations, and consequently, the Perception-Action Model. Instead, when task requirements are controlled for, we find similar behavioural performance in perception and action, suggesting that they are based on a common representation of object features in the visual system.