Coherence from the default mode network

coherenbrainThis post starts with the paper “Brains striving for coherence: Long-term cumulative plot formation in the default mode network,” authored by K. Tylén, P. Christensen, A. Roepstorff, T. Lund, S. Østergaard, and M. Donald. The paper appeared in NeuroImage 121 (2015) 106–114.

People are  capable of navigating and keeping track of all the parallel social activities of everyday life even when confronted with interruptions or changes in the environment. Tylen et al suggest that even though these situations present themselves in series of interrupted segments often scattered over huge time periods, they tend to constitute perfectly well-formed and coherent experiences in conscious memory. However, the underlying  mechanisms of such long-term integration is not well understood. While brain activity is generally traceable within the short time frame of working memory, these integrative processes last for minutes, hours or even days.

To examine this the researchers set up two experiments. The experiments looked at a task that requires such sustained attention–listening to a story.  I will leave out the detail of the experiments, but in summary, the first experiment involved subjects answering questions after having listened to stories that either were set up to allow for understanding the plot or not. The second experiment had people answer similar questions after listening to the stories while having their brains monitored by fMRI.

Experiment I found a cognitive dissociation between  comprehension of plot (such as characters’ motives, dispositions and key events) and incidental facts (such as details of the characters’ profession, age, brand of car, etc.) in narratives:  when episodic material allows for long-term integration in a coherent plot, we recall fewer factual details. However, when plot formation is challenged, we pay more attention to incidental facts. Tylen et al state that: “Participants’ behavior seems indicative of an implicit strategy to remember more casual details in the hope that subsequent events will allow them to retrospectively reconstruct the plot.”

Experiment II found a central role for the brain’s default mode network related to comprehension of coherent narratives while incoherent episodes rather activate the frontoparietal control network. Experiment II findings thus suggest a hierarchical cortical
organization with longer timescales of integration selectively recruiting areas of the default mode network.  In other studies, the default mode network has been found to be involved in the continuous updating of prospective mental models to guide attention and action in anticipatory ways. The default mode network comprises areas along the anterior and posterior midline, the lateral parietal cortex, prefrontal cortex, and the medial temporal lobe. It was originally found to activate when experimental participants were not focusing attention on the here-and-now immediacy of a task and was thus thought to reflect the ‘resting brain’ or ‘spontaneous, unconstrained thought.’  Findings now suggest that the default mode network flexibly couples with other networks to accomplish memory-related functional goals. For instance, it has been found that a network of areas in lateral prefrontal and parietal cortex, termed the frontoparietal control network, coactivate with the default mode network as a function of increased task demands.

I cannot decide whether or not these were clever experiments or not, but regardless it seems to provide evidence of neurological networks that match up with the networks in the parallel constraint satisfaction model. Though maybe not an exact match for intuition/automatic  or primary network of the PCS model, the default mode network seems quite close. The default mode network is whirring away grabbing coherence from the parallel social activities of humans, while intuition is whirring away running its simulations tweaking the numbers to reduce dissonance. Is there a difference? Then we can match up the frontoparietal control network with the deliberate/analysis network.  The frontoparietal control network is enlisted to seek out facts to try to create coherence as the facts unwind. The analytic network is similarly enlisted to get more information to feed the intuitive system to reduce dissonance or create coherence.

Tylen et al note that individual variability in participants’ default mode network (DMN) responses was predicted by their behavioral responses to the memory questions in the scanner. Participants who paid less attention to incidental facts in coherent stories than in incoherent stories had better fits to the model in temporal parts of the DMN, than those who attended more evenly to facts in both coherent and incoherent stories. According to the authors, this is a strong indication that modulations in DMN activity is indeed systematically linked to cumulative plot formation. I cannot help but think of Parameter P in the parallel constraint satisfaction model at this juncture. Parameter P allows PCS-DM to capture individual differences in the subjective sensitivity to differences in cue validities (see post Parameter P–Slowness Factor?). Low sensitivity is captured by low P. Maybe that would equate with low sensitivity to incidental facts. P captures sensitivity at the level of individuals, that is, it determines how an individual transforms explicitly provided or learned information about a cue’s predictive power (i.e., cue validity) into a weight. The suggestion of Glockner et al that P describes a core property of a psychological transformation process that precedes decision making  seems to match up with the conclusions of Tylen et al.

Back to the idea of the coherence seeking default mode network, Ken Hammond explained this well decades ago (post Judgments under Stress). Coherence constancy refers to a person’s ability to apply the same coherent principle to a problem, despite its various forms of display or presentation. Hammond suggested that the Gestaltists have shown over and over that we have the capacity of intuitive coherence through often perceiving what ought to be present rather than what is is present.  But analytical coherence must be taught.  Thus, the default mode network does not need a perfect story to create a coherent plot–that is what it does. If there is anything close to coherence, it uses bidirectional reasoning to create a coherent plot. But when the story is so disjointed that we see something wrong, we need to be taught how to make sense of those odd facts. Numeracy and probability understanding are helpful.

Hammond said:  “Common sense (robust flexibility) means engaging in as much analytical work as required, and in as much intuition as will suffice, because intuition is by far the easiest.”  Finding the right mix of rules and discretion varies, and successful cognition must be adapted to the structure of information in the environment. Now the neurological research is providing clear backing for some of Ken Hammond’s ideas and providing strong backing for the parallel constraint satisfaction model.