An Information Integration Approach to Perception of Causality Trough Continuous Behavioural Measures

The study of perceptual causality has known a regain in interest from the moment that continuous behavioural measures, and not just perceptual reports, were proposed. As one such measure, representational momentum (RM), was recently applied by T. Hubbard to Michotte’s launching paradigm. Among the several open issues regarding the use of RM to gauge perceptual causality there are the following two: 1) how the joint action of dynamic and kinematics variables is captured and integrated in RM; 2) the influence of specific response modalities. In this work “launcher-target” velocities, size of launcher (“implied mass”), gap at the collision, and distance travelled by the target were fully crossed in two information integration tasks, which demanded a localisation response either through a mouse cursor or with a pointer. A significant role of velocity, mass and gap was documented, with mass and velocity combining additively. Integration patterns were shown the same in both response modalities, although “pointing” appears as an advantageous response modality to address RM-attenuation effects. The finding of a systematic reduction in memory forward displacement (RM) for targets involved in collision settings similar to Michotte’s “launching” and “tool” paradigms (Hubbard et al, 2001; Hubbard & Ruppel, 2002; Hubbard & Favretto, 2003), promoted RM as a behavioural measure of causal perception. One major advantage of such a measure over perceptual reports is the circumstance that, by being continuous, it makes possible to address the issue of the integration of contributing factors to causal perception. Many such determinants have been studied in isolation (speed, gap, size, acceleration, shape, elasticity, friction, direction, etc.) but seldom in their joint action. One notable exception is the study by Schlottmann & Anderson (1993), who used continuous ratings of causality and naturalness to establish an integration model of phenomenal causality. Anderson’s IIT methodology (Anderson, 1981; 1982; 1996) has actually been offering a powerful framework to the use of continuous response variables across numerous fields. This work stems from the notion that RM offers new prospects for the use of IIT in the realm of causal perception and naïve physics. As a counterpart, further insights might be gained on the extent to which high-level cognitive processes are reflected in RM, a centre issue of the debate over the very nature of forward displacement bias (Hubbard, 2005; 2006). One particular topic of integration concerns the combined action of mass (as implied by object’s size) and speed on RM. Differently from physical momentum, defined as mass times speed, the mass of a horizontally moving object presented in isolation produces no effect on RM (a non-integration result). The situation is different for vertical movements, along the axis of implied gravity, where mass starts acting as implied weight – a force (Hubbard, 1997). This points to the potential usefulness of distinguishing between typical kinematics variables such as speed and more dynamics-associated variables such as “mass”, whose effects on RM might require settings which turn it into a force enabling a specific result. In the present work, the integration of implied mass and velocity is studied in a collisions setting (launching displays) where implied mass can become part of a “force of impact”. The size of a launcher object is manipulated jointly with pairs of “launcher-target” velocities (keeping a constant target/launcher speed ratio of 1/3). Other kinematics and spatial variables commonly considered in studies of launching and in RM studies were also included (gap at collision, target’s travelled distance). In line with IIT methodology, all variables were fully crossed in factorial integration tasks, demanding a localisation response. As a supplementary betweensubjects factor, two kinds of localisation response, either indirectly through a mouse cursor, or directly through a pointer, were used.