Opinion - The relevance of behavioural measures for functional-imaging studies of cognition

The psychological structure of cognition is often inferred from conjoint measures of behaviour (such as reaction time) and brain activation (such as cerebral blood flow). In many experiments these measures produce divergent results. One example is where a significant pattern of brain activation occurs without a corresponding change in overt behaviour. In such circumstances, can cognitive theory be informed from brain-activation data alone? Or, given the more established link between behaviour and cognition, is behavioural corroboration necessary?