Abstract
Components of neuropsychological tests commonly employed by neuroimagers in developing a metric for Theory of Mind (ToM) research are currently extremely mixed. Tests commonly include factors which may be separate from the core ToM component such as emotion, morality and verbal skills. We explore these confounds, and hypothesize that the discrepant conclusions that have arisen in neuroimaging ToM research reflect the degree to which a task is confounded by these factors. To test this hypothesis we conduct a coordinate based activation likelihood estimations (ALE) meta-analysis of neuroimaging studies. The results demonstrate divergences in neuroimaging results are predicted by divergences in the use of verbal or non-verbal tasks. In response to this finding we systematically describe the main dimensions that should be excluded and included in an ideal ToM task drawing on the wide literature surveyed. We argue, good tasks should rely on minimum verbal content; avoid excessive empathic and moral reactions; include human-like figures executing purposeful behaviors; present several standardized phases, with increased levels of difficulty; and incorporate a clear scoring system, so that performance can be categorized. In order to try to accomplish these goals, we introduce the new ‘Inverted Comic Strips Task’.