Different studies have suggested that the majority of self-talk during exercise is either positive or neutral in character. The majority of 'thoughts' during low-intensity exercise have been described as being dissociative conversational chatter. However, with increasing exercise intensity, there is a greater percentage of associative and motivational thoughts, which includes thoughts about feeling and affect, body monitoring, command, instruction and pace monitoring. It has been suggested that self-talk is necessary for creating a time 'wedge' between the activity described by the self-talk, and the self-talk itself. The information redundancy created by this time-wedge allows the capacity for reflection about what is occurring, and self-awareness of the part played by the individual themselves in the activity being performed. Self-talk may be a discussion between a singular 'I' and a singular 'me', or may be a multi-party dialogue. There are anatomical correlates to self-talk, with neural activity in a number of brain areas related to the occurrence of both overt and subvocal self-talk, particularly in Broca's region in the left fontal cortex, and Wernicke's region in the left posterior superior temporal cortex. Whether specific training of self-talk can improve performance is controversial, although recent studies have suggested that task-specific self-talk appears to have a beneficial effect on physical performance. Further studies are required to assess the ability of physical or mental training to modify self-talk in a beneficial and permanent manner, and whether these changes affect an individual's exercise performance and sense of self.