Correlates of Complex Imitation Fidelity to Patterns of BOLD Signal Generated During An fMRI Study of Simple Imitation

Background:

Imitation is a method wherein we learn actions by seeing others perform them, and consequently replicate these actions to make them our own. This social learning is likely to be a crucial aspect of our social cognitive development. Research shows that, although they can imitate, people with autism perform less well on certain tests of imitation. It is therefore important to identify the factors that determine between-subject variability in imitation fidelity. Whilst a significant amount of research has explored the broad neural substrate of imitation, little is known about the role of individual differences, and the brain areas that might predict whether people are good or poor at imitation.

Objectives:

To address the issue, our study sought to determine where variable patterns of BOLD signal, generated whilst performing a simple imitation task, predicted performance on a demanding imitation task conducted outside the scanner. The specific objectives were:

a)             To obtain objective measures of performance from a complex imitation task

b)             To obtain measures of BOLD correlates of neural activity on a goal-directed action imitation and observation task

c)             To see the influence of between-subject variability in imitation fidelity on BOLD activation

Methods:

            15 participants were recruited to perform fMRI and behavioural tasks. Inside the MRI-scanner, participants were instructed to imitate different grips on a handle, or to simply watch someone or a circle moving the handle. Outside the scanner, participants were given a touchscreen-laptop with stylus and instructed to first look at a video of a model drawing a shape using the same apparatus, and then to imitate that drawing, matching the model in speed, size and shape. There were 45 different drawings in total. The speeds and path lengths of these drawings were compared to those of the model, resulting in individual drawing speed and size measures of accuracy, precision and error in imitation fidelity. We then tested for correlations between these behavioural measures and BOLD signal during simple imitation.

Results:

            Activity in the ‘Imitate’ BOLD-contrast correlated negatively with accuracy for path length in left postcentral parietal lobe and with speed in right ventromedial frontal gyrus and anterior parietal lobe. The gradient of change of participant’s speed of action in relation to that of the model correlated negatively with Imitation-related activity in left medial frontal gyrus, bilateral insula and cerebellum, right intraparietal sulcus and thalamus. Error for speed correlated positively with Imitate in left posterior cingulate and intraparietal sulcus.

Conclusions:

            Greater skill in performing a demanding imitation task outside the scanner was associated with less BOLD activity in brain areas thought to serve imitation during a simple imitation task. We suggest that this reflects better adaptation of these areas to the function of imitation. Additionally, different measures of imitation fidelity corresponded to different activation patterns, suggesting that various aspects of imitation rely differentially on a range of functions that may include action representation and feedback sensitivity.