Using goal- and grip-related information for understanding the correctness of other's actions: an ERP study

Abstract

Detecting errors in other's actions is of pivotal importance for joint action, competitive behavior and observational learning. Although many studies have focused on the neural mechanisms involved in detecting low-level errors, relatively little is known about error-detection in everyday situations. The present study aimed to identify the functional and neural mechanisms whereby we understand the correctness of other's actions involving well-known objects (e.g. pouring coffee in a cup). Participants observed action sequences in which the correctness of the object grasped and the grip applied to a pair of objects were independently manipulated. Observation of object violations (e.g. grasping the empty cup instead of the coffee pot) resulted in a stronger P3-effect than observation of grip errors (e.g. grasping the coffee pot at the upper part instead of the handle), likely reflecting a reorienting response, directing attention to the relevant location. Following the P3-effect, a parietal slow wave positivity was observed that persisted for grip-errors, likely reflecting the detection of an incorrect hand-object interaction. These findings provide new insight in the functional significance of the neurophysiological markers associated with the observation of incorrect actions and suggest that the P3-effect and the subsequent parietal slow wave positivity may reflect the detection of errors at different levels in the action hierarchy. Thereby this study elucidates the cognitive processes that support the detection of action violations in the selection of objects and grips.

Figure 1. Example stimuli used in the experiment.

For each object pair an action sequence was constructed, consisting of 3 action snapshots (left column, middle column and right column). The correctness of the object grasped and the grip applied to the object were independently manipulated, resulting in action sequences representing an actor (1) grasping the correct object with a correct grip (upper row), (2) grasping the correct object with an incorrect grip (2^nd row), (3) grasping the incorrect object with a correct grip (3^rd row) or (4) grasping the incorrect object with an incorrect grip (bottom row).

Figure 2. Schematic overview of a trial sequence.

Each trial started with a fixation cross, after which the action sequence was presented, consisting of 3 action snapshots. Following the action sequence either an object question (left side) or a grip question (right side) could be presented. Mapping of the response buttons (left/right) varied between trials and was presented below the question. After the subject responded a blank screen was presented.

Figure 3. Behavioral results.

Error rates in response to object questions (left graph) or to grip questions (right graph). Bars on the left represent responses to action sequences representing grasping of the correct object, bars on the right represent responses to action sequences representing grasping of the incorrect object. Light bars represent responses to action sequences representing a correct grip, dark bars represent responses to action sequences representing an incorrect grip.

Figure 4. ERPs relative to the onset of the 2^nd picture.

ERPs relative to the onset of the 2^nd picture for a selection of central electrodes. Topographical plots represent the difference between Incorrect and Correct Objects (upper panel) and the difference between Incorrect and Correct Grips (middle panel. The lower panel reflects the t-values for the comparison between the Object Effect (Incorrect – Correct Object) and the Grip Effect (Incorrect – Correct Grip). The critical t-values are marked in red and a positive t-value reflects a stronger effect of Object than of Grip and a negative t-value reflects a stronger effect of Grip than of Object.