Where do we grasp objects? - An experimental verification of the Selective Attention for Action Model (SAAM)

Abstract

Classically, visual attention is assumed to be influenced by visual properties of objects, e.g. as assessed in visual search tasks. However, recent experimental evidence suggests that visual attention is also guided by action-related properties of objects ("affordances",[1,2]), e.g. the handle of a cup affords grasping the cup; therefore attention is drawn towards the handle (see [3] for example). In a first step towards modelling this interaction between attention and action, we implemented the Selective Attention for Action model (SAAM). The design of SAAM is based on the Selective Attention for Identification model (SAIM [4]). For instance, we also followed a soft-constraint satisfaction approach in a connectionist framework. However, SAAM's selection process is guided by locations within objects suitable for grasping them whereas SAIM selects objects based on their visual properties. In order to implement SAAM's selection mechanism two sets of constraints were implemented. The first set of constraints took into account the anatomy of the hand, e.g. maximal possible distances between fingers. The second set of constraints (geometrical constraints) considered suitable contact points on objects by using simple edge detectors. At first, we demonstrate here that SAAM can successfully mimic human behaviour by comparing simulated contact points with experimental data. Secondly, we show that SAAM simulates affordance-guided attentional behaviour as it successfully generates contact points for only one object in two-object images. © 2009 Springer Berlin Heidelberg.