Monitoring the mind: clues for a link between meta cognition and self generated thought

It is a relatively common experience to lose track of what one is doing: We may stop following what someone is saying during conversation, enter a room and realise we have forgotten why we came in, or lose the thread of our own thoughts leaving us with a sense that we had reached a moment of insight that is now lost forever. One important influence on making sure that we can stay on target to achieve our goals is the capacity for meta-cognition, or the ability to accurately assess our own cognitive experience. Meta cognition is important because it allows us the opportunity to correct for errors if and when they occur. I have recently become interested in this capacity for accurately assessing the contents of thought and along with two different groups of collaborators have begun to explore its neural basis.

We were interested in whether meta-cognition is a unitary process or instead can take multiple forms. In a study published this year in the Journal of Neuroscience and led by Benjamin Baird, a PhD student at the University of California, Santa Barbara, and Daniel Margulies and Chris Gorgelowski from the Max Planck Institute in Leipzig, we explored whether the capacity for meta-cognition for information from memory was related to the capacity to reflect on perceptual input. We measured participant’s meta-cognitive accuracy for how they performed on two tasks – one which depended upon immediate meta cognitive accuracy for rapidly presented perceptual information and a second which depended upon the ability to reflect on task performance regarding information presented at a longer time scale (in this case words presented about quarter of an hour earlier). In addition to this behavioural investigation we used resting state functional connectivity to understand whether these different capacities were similar or different at the neural level.

The first element of our results that was surprising was that the correlation between peoples meta-cognitive accuracy was so low that across our 54 participants the two tasks were not significantly associated. This suggests that at a behavioural level these different forms of meta-cognition are not a single uniform trait. Instead these capacities are sufficiently different that a person’s performance one task is relatively uninformative of their performance on the other.

We also found dissociation at the neural level. We looked at how our participant’s performance on these tasks was related to the neural changes we observed when our participants were left to rest in a functional magnetic resonance imaging (fMRI) scanner. We found that the different forms of meta-cognition were associated with distinct neural networks. People who had better meta-cognitive accuracy for perception exhibited greater connectivity between a small region on the lateral surface of the frontal cortex and regions of the thalamus, anterior cingulate cortex and the putamen. By contrast people who had better meta-cognition for information from memory had greater connectivity between a medial region of the frontal cortex, a region of the posterior cingulate cortex (pCC) and the inferior parietal sulcus (IPS). Together with this region of the medial frontal cortex, the pCC and IPS form a network known as the default mode network that is thought to be important in self-generated thoughts (such as mind wandering and daydreaming).

The neural data, therefore, extended our behavioural result by demonstrating that meta-cognitive accuracy for perception was dependent upon different a neural network than was meta-cognitive accuracy of information from memory. Importantly whereas the behavioural correlation suggested that the two processes were uncorrelated (a null result), our neuroimaging results provided evidence that they were indeed different. Here is the link to our paper if you would like to read more (

A second study, led by Micah Allen a post doctoral student then at the University of Aarhus, Denmark, and Antoine Lutz, a researcher in Lyon, France, and published in the journal Frontiers in Human Neuroscience, was concerned with understanding how task unrelated thought relates to meta-cognitive accuracy for on-going performance.  In this experiment we asked participants to perform a demanding cognitive task in which task unrelated thought was measured; We also asked participants to recognise if and when they made an error on the task. This provided a measure of meta cognition of on going task performance. Testing took place in an fMRI scanner and participants were given a financial motivation to perform the task properly.

We were interested in exploring how variability in task unrelated thought would be related to awareness of lapses in performance (e.g. meta cognitive accuracy for actions based on immediate information). In simple terms being on task is sometimes a good thing (for example when operating heavy machinery) and at other times being off task can be beneficial (like when preparing for a business meeting on the commute to work) and we wanted to understand how people might manage this trade-off. Our reasoning was that task unrelated thoughts could be adaptive if the capacity for self generated and perceptually generated experience was balanced. We operationalised this balance between task related and task unrelated thoughts as variability in their tendency to be off task across the duration of the experiment.

In neural terms we found that noticing an error – a meta cognitive judgement regarding on going task performance – elicited neural activation in a similar network as was predictive of meta cognitive accuracy for perceptual judgements in our prior paper. As before this pattern involved the anterior cingulate, as well as the putamen and thalamus.

Replicating numerous other studies, we also found that people who engaged in more task unrelated thoughts did worse on the primary task. This deleterious impact of task unrelated thinking is one of the major costs of self-generated thought. However, we also found that variance in task unrelated thought was predictive of a greater capacity to recognise a lapse when and if it occurs. Critically, variance in task unrelated thoughts was not associated with better performance on a task, but it was associated with a better capacity to recognise an error when it occurred. Thus meta cognitive accuracy for performance was higher for people who were neither task focused, nor off tasks but who instead tended to vary between both states throughout the course of the experiment. This could mean that a critical output of meta cognitive monitoring of performance is the promotion of variations in the focus of conscious attention. You can read our paper here (

Altogether these two studies suggest that meta-cognition is deeply related to the experience of self-generated thought. Our resting state study suggests that meta-cognition for long term semantic information depends upon the default mode network. This core system of brain regions is known to be important for experiences such as mind-wandering or daydreaming. By contrast, a network involving the anterior cingulate, putamen and thalamus seems important in meta cognition for perceptual information and is related to an individuals capacity to balance their minds tendency to self-generate thoughts unrelated to the task in hand with those that depend on a more detailed focus on on-going actions.

It seems possible that meta-cognitive processes are important to experiences such as daydreaming, or mind-wandering, for at least two reasons. First, the capacity for the meta cognitive analysis of information from memory may account for our ability to successfully engage in behaviours that depend on our past experience. This may arise from the activity of the default mode network and it could be that it is this capacity for reflecting on memorial information that helps us navigate the complex world in which we exist in a creative manner. Second, the capacity to recognise that on going action may be in jeopardy, perhaps because attention has switched to self-generated thoughts, is a tendency of people who balance task related and task unrelated thoughts. Functioning in combination these different forms of meta cognition may help an individual integrate two modes of cognition – that have often be conceived of as being fundamentally different – in an adaptive fashion.

Micah Allen is now a post doctoral researcher at the Institute of Cognitive Neuroscience at University College London.

This entry was posted in Uncategorized and tagged , , , , , . Bookmark the permalink.