Understanding the Goals of Everyday Instrumental Actions Is Primarily Linked to Object, Not Motor-Kinematic, Information: Evidence from fMRI

Abstract

Prior research conceptualised action understanding primarily as a kinematic matching of observed actions to own motor representations but has ignored the role of object information. The current study utilized fMRI to identify (a) regions uniquely involved in encoding the goal of others' actions, and (b) to test whether these goal understanding processes draw more strongly on regions involved in encoding object semantics or movement kinematics. Participants watched sequences of instrumental actions while attending to either the actions' goal (goal task), the movements performed (movement task) or the objects used (object task). The results confirmed, first, a unique role of the inferior frontal gyrus, middle temporal gyrus and medial frontal gyrus in action goal understanding. Second, they show for the first time that activation in the goal task overlaps directly with object- but not movement-related activation. Moreover, subsequent parametric analyses revealed that movement-related regions become activated only when goals are unclear, or observers have little action experience. In contrast to motor theories of action understanding, these data suggest that objects-rather than movement kinematics-carry the key information about others' actions. Kinematic information is additionally recruited when goals are ambiguous or unfamiliar.

Fig 1. Behavioural task.

Figure 1 presents a schematic image of the behavioural task. The left hand side of the image illustrates the (fully counterbalanced) sequence of the mini-blocks/tasks across a run (passive viewing baseline, object task, movement task and goal task). On the right, an example of an experimental mini block for each condition is depicted. Note that the image with a red border within each mini-block illustrates a repetition for that condition (e.g. a repeated object, movement or goal).

Fig 2. Unique goal activations & pairwise comparisons of goal, movement and object tasks.

Panel A shows unique activation in the goal task (green), as well as activation shared between goal task and object task (blue). No activations were shared with the movement task. Panel B to D show pairwise comparisons of the three tasks. Panel B: Object task (blue) vs. Movement task (red). Panel C: Goal task (green) vs. Movement task (red). Panel D: Goal task (green) vs. Object task (blue). All activations thresholded at p < 0.005 and whole-brain corrected to a familywise error p < 0.05.

Fig 3. Parametric activations.

Shows brain regions showing associations with the participants’ subjective action ratings. Top panel: negative associations with the apparentness of the action goals. Lower panel: negative associations with the amount of prior sensorimotor experience with the actions. All activations thresholded at p < 0.005 and whole-brain corrected to a familywise error p < 0.05.

Fig 4. A simple model of everyday action understanding.

Potential goals are initially identified on the basis of object information in inferior frontal and middle temporal areas (Step 1). Parietal-premotor motor-representation regions verify this initial interpretation or provide additional information (Step 2). If a goal is identified, associated mental states can in turn become activated—via medial prefrontal areas—and integrated with the situational context or prior knowledge about the person (Step 3).