Dissociating action-effect activation and effect-based response selection

Action control costs in task selection: Agents avoid actions with incompatible movement and effect features

When a movement triggers effects with incompatible features, conflict between action and effect features creates costs for action planning and initiation. We investigated whether such action control costs also factor into action choices in terms of the principle of least effort. Participants completed a reaction-time experiment, where they were instructed to perform left and right mouse swipes in response to directional cues presented on the screen. Participants could select between two action options on each trial: Depending on which part of the screen (upper or lower) the action was performed in, the swipe resulted in a visual stimulus moving in the same (compatible) or in the opposite (incompatible) direction as the mouse. Incompatible action-effect mappings did indeed incur action control costs. In accordance with effort avoidance, the proportion of compatible choices was significantly above chance level, suggesting that action selection and initiation costs factor into participants preferences. Interestingly, however, participants' choice tendencies were not predicted by the actual increase in action-initiation costs in the incompatible condition. This indicates that effort-related decisions are not simply based on monitoring performance in the actual task, but they are also influenced by preestablished notions of action-planning costs.

Spatial action-effect binding depends on type of action-effect transformation

Spatial action-effect binding denotes the mutual attraction between the perceived position of an effector (e.g., one's own hand) and a distal object that is controlled by this effector. Such spatial binding can be construed as an implicit measure of object ownership, thus the belonging of a controlled object to the own body. The current study investigated how different transformations of hand movements (body-internal action component) into movements of a visual object (body-external action component) affect spatial action-effect binding, and thus implicit object ownership. In brief, participants had to bring a cursor on the computer screen into a predefined target position by moving their occluded hand on a tablet and had to estimate their final hand position. In Experiment 1, we found a significantly lower drift of the proprioceptive position of the hand towards the visual object when hand movements were transformed into laterally inverted cursor movements, rather than cursor movements in the same direction. Experiment 2 showed that this reduction reflected an elimination of spatial action-effect binding in the inverted condition. The results are discussed with respect to the prerequisites for an experience of ownership over artificial, noncorporeal objects. Our results show that predictability of an object movement alone is not a sufficient condition for ownership because, depending on the type of transformation, integration of the effector and a distal object can be fully abolished even under conditions of full controllability.

The interplay of predictive and postdictive components of experienced selfhood

Objects that we affect by our body movements can be experienced as being controlled by (agency) and belonging to the own body (ownership). Such impressions of minimal selfhood arise when objects move as predicted prior to the action (predictive component). But they can also arise when otherwise unpredictable object movements turn out to be consistent with (e.g. spatially compatible to) preceding actions (postdictive component). Here we studied how the impact of postdictive components of inferred minimal selfhood in terms of action-object compatibility is shaped by different levels of predictability of these object movements. We found that compatibility between actions and object movements, and to a lesser extent predictability of object movements, affected reported agency while only compatibility affected reported ownership. Importantly, predictive and postdictive factors influenced these measures in an independent manner. We discuss these results against the background of models that assume multiple components of experienced minimal selfhood.