Explaining distortions in metacognition with an attractor network model of decision uncertainty

Metacognition is the ability to reflect on, and evaluate, our cognition and behaviour. Distortions in metacognition are common in mental health disorders, though the neural underpinnings of such dysfunction are unknown. One reason for this is that models of key components of metacognition, such as decision confidence, are generally specified at an algorithmic or process level. While such models can be used to relate brain function to psychopathology, they are difficult to map to a neurobiological mechanism. Here, we develop a biologically-plausible model of decision uncertainty in an attempt to bridge this gap. We first relate the model’s uncertainty in perceptual decisions to standard metrics of metacognition, namely mean confidence level (bias) and the accuracy of metacognitive judgments (sensitivity). We show that dissociable shifts in metacognition are associated with isolated disturbances at higher-order levels of a circuit associated with self-monitoring, akin to neuropsychological findings that highlight the detrimental effect of prefrontal brain lesions on metacognitive performance. Notably, we are able to account for empirical confidence judgements by fitting the parameters of our biophysical model to first-order performance data, specifically choice and response times. Lastly, in a reanalysis of existing data we show that self-reported mental health symptoms relate to disturbances in an uncertainty-monitoring component of the network. By bridging a gap between a biologically-plausible model of confidence formation and observed disturbances of metacognition in mental health disorders we provide a first step towards mapping theoretical constructs of metacognition onto dynamical models of decision uncertainty. In doing so, we provide a computational framework for modelling metacognitive performance in settings where access to explicit confidence reports is not possible.     

Simulating Fiction: Individual Differences in Literature Comprehension Revealed with fMRI

When we read literary fiction, we are transported to fictional places, and we feel and think along with the characters. Despite the importance of narrative in adult life and during development, the neurocognitive mechanisms underlying fiction comprehension are unclear. We used functional magnetic resonance imaging (fMRI) to investigate how individuals differently employ neural networks important for understanding others’ beliefs and intentions (mentalizing), and for sensori-motor simulation while listening to excerpts from literary novels. Localizer tasks were used to localize both the cortical motor network and the mentalizing network in participants after they listened to excerpts from literary novels. Results show that participants who had high activation in anterior medial prefrontal cortex (aMPFC; part of the mentalizing network) when listening to mentalizing content of literary fiction, had lower motor cortex activity when they listened to action-related content of the story, and vice versa. This qualifies how people differ in their engagement with fiction: some people are mostly drawn into a story by mentalizing about the thoughts and beliefs of others, whereas others engage in literature by simulating more concrete events such as actions. This study provides on-line neural evidence for the existence of qualitatively different styles of moving into literary worlds, and adds to a growing body of literature showing the potential to study narrative comprehension with neuroimaging methods.     

Differentiating Self-Projection from Simulation during Mentalizing: Evidence from fMRI

We asked participants to predict which of two colors a similar other (student) and a dissimilar other (retiree) likes better. We manipulated if color-pairs were two hues from the same color-category (e.g. green) or two conceptually different colors (e.g. green versus blue). In the former case, the mental state that has to be represented (i.e., the percept of two different hues of green) is predominantly non-conceptual or phenomenal in nature, which should promote mental simulation as a strategy for mentalizing. In the latter case, the mental state (i.e. the percept of green versus blue) can be captured in thought by concepts, which facilitates the use of theories for mentalizing. In line with the self-projection hypothesis, we found that cortical midline areas including vmPFC / orbitofrontal cortex and precuneus were preferentially activated for mentalizing about a similar other. However, activation was not affected by the nature of the color-difference, suggesting that self-projection subsumes simulation-like processes but is not limited to them. This indicates that self-projection is a universal strategy applied in different contexts—irrespective of the availability of theories for mentalizing. Along with midline activations linked to self-projection, we also observed activation in right lateral frontal and dorsal parietal areas showing a theory-like pattern. Taken together, this shows that mentalizing does not operate based on simulation or theory, but that both strategies are used concurrently to predict the choices of others.     

Mind blindness and the brain in autism.

Experimental evidence shows that the inability to attribute mental states, such as desires and beliefs, to self and others (mentalizing) explains the social and communication impairments of individuals with autism. Brain imaging studies in normal volunteers highlight a circumscribed network that is active during mentalizing and links medial prefrontal regions with posterior superior temporal sulcus and temporal poles. The brain abnormality that results in mentalizing failure in autism may involve weak connections between components of this system.     

Encoding predicted outcome and acquired value in orbitofrontal cortex during cue sampling depends upon input from basolateral amygdala

Certain goal-directed behaviors depend critically upon interactions between orbitofrontal cortex (OFC) and basolateral amygdala (ABL). Here we describe direct neurophysiological evidence of this cooperative function. We recorded from OFC in intact and ABL-lesioned rats learning odor discrimination problems. As rats learned these problems, we found that lesioned rats exhibited marked changes in the information represented in OFC during odor cue sampling. Lesioned rats had fewer cue-selective neurons in OFC after learning; the cue-selective population in lesioned rats did not include neurons that were also responsive in anticipation of the predicted outcome; and the cue-activated representations that remained in lesioned rats were less associative and more often bound to cue identity. The results provide a neural substrate for representing acquired value and features of the predicted outcome during cue sampling, disruption of which could account for deficits in goal-directed behavior after damage to this system.     

The Social Bayesian Brain: Does Mentalizing Make a Difference When We Learn?

When it comes to interpreting others'' behaviour, we almost irrepressibly engage in the attribution of mental states (beliefs, emotions…). Such "mentalizing" can become very sophisticated, eventually endowing us with highly adaptive skills such as convincing, teaching or deceiving. Here, sophistication can be captured in terms of the depth of our recursive beliefs, as in "I think that you think that I think…" In this work, we test whether such sophisticated recursive beliefs subtend learning in the context of social interaction. We asked participants to play repeated games against artificial (Bayesian) mentalizing agents, which differ in their sophistication. Critically, we made people believe either that they were playing against each other, or that they were gambling like in a casino. Although both framings are similarly deceiving, participants win against the artificial (sophisticated) mentalizing agents in the social framing of the task, and lose in the non-social framing. Moreover, we find that participants'' choice sequences are best explained by sophisticated mentalizing Bayesian learning models only in the social framing. This study is the first demonstration of the added-value of mentalizing on learning in the context of repeated social interactions. Importantly, our results show that we would not be able to decipher intentional behaviour without a priori attributing mental states to others.     

Dissociable medial prefrontal contributions to judgments of similar and dissimilar others

Human social interaction requires the recognition that other people are governed by the same types of mental states-beliefs, desires, intentions-that guide one's own behavior. We used functional neuroimaging to examine how perceivers make mental state inferences when such self-other overlap can be assumed (when the other is similar to oneself) and when it cannot (when the other is dissimilar from oneself). We observed a double dissociation such that mentalizing about a similar other engaged a region of ventral mPFC linked to self-referential thought, whereas mentalizing about a dissimilar other engaged a more dorsal subregion of mPFC. The overlap between judgments of self and similar others suggests the plausibility of "simulation" accounts of social cognition, which posit that perceivers can use knowledge about themselves to infer the mental states of others.     

The neural bases of social intention understanding: the role of interaction goals

Decoding others' intentions is a crucial aspect of social cognition. Neuroimaging studies suggest that inferring immediate goals engages the neural system for action understanding (i.e. mirror system), while the decoding of long-term intentions requires the system subserving the attribution of mental states (i.e. mentalizing). A controversial issue, stimulated by recent inconsistent results, concerns whether the two systems are concurrently vs. exclusively involved in intention understanding. This issue is particularly relevant in the case of social interactions, whose processing has been mostly, but not uncontroversially, associated with the mentalizing system. We tested the alternative hypothesis that the relative contribution of the two systems in intention understanding may also depend on the shared goal of interacting agents. To this purpose, 27 participants observed social interactions differing in their cooperative vs. affective shared goal during functional-Magnetic-Resonance-Imaging. The processing of both types of interactions activated the right temporo-parietal junction involved in mentalizing on action goals. Additionally, whole-brain and regions-of-interest analyses showed that the action understanding system (inferior prefrontal-parietal cortex) was more strongly activated by cooperative interactions, while the mentalizing-proper system (medial prefrontal cortex) was more strongly engaged by affective interactions. These differences were modulated by individual differences in empathizing. Both systems can thus be involved in understanding social intentions, with a relative weighting depending on the specific shared goal of the interaction.     

The Human Factor: Behavioral and Neural Correlates of Humanized Perception in Moral Decision Making

The extent to which people regard others as full-blown individuals with mental states (“humanization”) seems crucial for their prosocial motivation towards them. Previous research has shown that decisions about moral dilemmas in which one person can be sacrificed to save multiple others do not consistently follow utilitarian principles. We hypothesized that this behavior can be explained by the potential victim’s perceived humanness and an ensuing increase in vicarious emotions and emotional conflict during decision making. Using fMRI, we assessed neural activity underlying moral decisions that affected fictitious persons that had or had not been experimentally humanized. In implicit priming trials, participants either engaged in mentalizing about these persons (Humanized condition) or not (Neutral condition). In subsequent moral dilemmas, participants had to decide about sacrificing these persons’ lives in order to save the lives of numerous others. Humanized persons were sacrificed less often, and the activation pattern during decisions about them indicated increased negative affect, emotional conflict, vicarious emotions, and behavioral control (pgACC/mOFC, anterior insula/IFG, aMCC and precuneus/PCC). Besides, we found enhanced effective connectivity between aMCC and anterior insula, which suggests increased emotion regulation during decisions affecting humanized victims. These findings highlight the importance of others’ perceived humanness for prosocial behavior - with aversive affect and other-related concern when imagining harming more “human-like” persons acting against purely utilitarian decisions.     

The role of metacognition in human social interactions

Metacognition concerns the processes by which we monitor and control our own cognitive processes. It can also be applied to others, in which case it is known as mentalizing. Both kinds of metacognition have implicit and explicit forms, where implicit means automatic and without awareness. Implicit metacognition enables us to adopt a we-mode, through which we automatically take account of the knowledge and intentions of others. Adoption of this mode enhances joint action. Explicit metacognition enables us to reflect on and justify our behaviour to others. However, access to the underlying processes is very limited for both self and others and our reports on our own and others' intentions can be very inaccurate. On the other hand, recent experiments have shown that, through discussions of our perceptual experiences with others, we can detect sensory signals more accurately, even in the absence of objective feedback. Through our willingness to discuss with others the reasons for our actions and perceptions, we overcome our lack of direct access to the underlying cognitive processes. This creates the potential for us to build more accurate accounts of the world and of ourselves. I suggest, therefore, that explicit metacognition is a uniquely human ability that has evolved through its enhancement of collaborative decision-making.     

In the eye of the beholder: visual experience and categories in the human brain

How does experience change representations of visual objects in the brain? Do cortical object representations reflect category membership? In this issue of Neuron, Jiang et al. show that category training leads to sharpening of neural responses in high-level visual cortex; in contrast, category boundaries may be represented only in prefrontal cortex.     

A computational framework for the study of confidence in humans and animals

Confidence judgements, self-assessments about the quality of a subject's knowledge, are considered a central example of metacognition. Prima facie, introspection and self-report appear the only way to access the subjective sense of confidence or uncertainty. Contrary to this notion, overt behavioural measures can be used to study confidence judgements by animals trained in decision-making tasks with perceptual or mnemonic uncertainty. Here, we suggest that a computational approach can clarify the issues involved in interpreting these tasks and provide a much needed springboard for advancing the scientific understanding of confidence. We first review relevant theories of probabilistic inference and decision-making. We then critically discuss behavioural tasks employed to measure confidence in animals and show how quantitative models can help to constrain the computational strategies underlying confidence-reporting behaviours. In our view, post-decision wagering tasks with continuous measures of confidence appear to offer the best available metrics of confidence. Since behavioural reports alone provide a limited window into mechanism, we argue that progress calls for measuring the neural representations and identifying the computations underlying confidence reports. We present a case study using such a computational approach to study the neural correlates of decision confidence in rats. This work shows that confidence assessments may be considered higher order, but can be generated using elementary neural computations that are available to a wide range of species. Finally, we discuss the relationship of confidence judgements to the wider behavioural uses of confidence and uncertainty.     

Neural Basis of Psychological Growth following Adverse Experiences: A Resting-State Functional MRI Study

Over the past decade, research on the aftereffects of stressful or traumatic events has emphasized the negative outcomes from these experiences. However, the positive outcomes deriving from adversity are increasingly being examined, and such positive changes are described as posttraumatic growth (PTG). To investigate the relationship between basal whole-brain functional connectivity and PTG, we employed resting-state functional magnetic resonance imaging and analyzed the neural networks using independent component analysis in a sample of 33 healthy controls. Correlations were calculated between the network connectivity strength and the Posttraumatic Growth Inventory (PTGI) score. There were positive associations between the PTGI scores and brain activation in the rostral prefrontal cortex and superior parietal lobule (SPL) within the left central executive network (CEN) (respectively, r = 0.41, p < 0.001; r = 0.49, p < 0.001). Individuals with higher psychological growth following adverse experiences had stronger activation in prospective or working memory areas within the executive function network than did individuals with lower psychological growth (r = 0.40, p < 0.001). Moreover, we found that individuals with higher PTG demonstrated stronger connectivity between the SPL and supramarginal gyrus (SMG). The SMG is one of the brain regions associated with the ability to reason about the mental states of others, otherwise known as mentalizing. These findings suggest that individuals with higher psychological growth may have stronger functional connectivity between memory functions within the CEN and social functioning in the SMG, and that their better sociality may result from using more memory for mentalizing during their daily social interactions.     

Functional role of the ventrolateral prefrontal cortex in decision making

To make deliberate decisions, we have to utilize detailed information about the environment and our internal states. The ventral visual pathway provides detailed information on object identity, including color and shape, to the ventrolateral prefrontal cortex (VLPFC). The VLPFC also receives motivational and emotional information from the orbitofrontal cortex and subcortical areas, and computes the behavioral significance of external events; this information can be used for elaborate decision making or design of goal-directed behavior. In this review, we discuss recent advances that are revealing the neural mechanisms that underlie the coding of behavioral significance in the VLPFC, and the functional roles of these mechanisms in decision making and action programming in the brain.     

Abnormalities in the awareness and control of action

Much of the functioning of the motor system occurs without awareness. Nevertheless, we are aware of some aspects of the current state of the system and we can prepare and make movements in the imagination. These mental representations of the actual and possible states of the system are based on two sources: sensory signals from skin and muscles, and the stream of motor commands that have been issued to the system. Damage to the neural substrates of the motor system can lead to abnormalities in the awareness of action as well as defects in the control of action. We provide a framework for understanding how these various abnormalities of awareness can arise. Patients with phantom limbs or with anosognosia experience the illusion that they can move their limbs. We suggest that these representations of movement are based on streams of motor commands rather than sensory signals. Patients with utilization behaviour or with delusions of control can no longer properly link their intentions to their actions. In these cases the impairment lies in the representation of intended movements. The location of the neural damage associated with these disorders suggests that representations of the current and predicted state of the motor system are in parietal cortex, while representations of intended actions are found in prefrontal and premotor cortex.     

The role of social cognition in decision making

Successful decision making in a social setting depends on our ability to understand the intentions, emotions and beliefs of others. The mirror system allows us to understand other people's motor actions and action intentions. 'Empathy' allows us to understand and share emotions and sensations with others. 'Theory of mind' allows us to understand more abstract concepts such as beliefs or wishes in others. In all these cases, evidence has accumulated that we use the specific neural networks engaged in processing mental states in ourselves to understand the same mental states in others. However, the magnitude of the brain activity in these shared networks is modulated by contextual appraisal of the situation or the other person. An important feature of decision making in a social setting concerns the interaction of reason and emotion. We consider four domains where such interactions occur: our sense of fairness, altruistic punishment, trust and framing effects. In these cases, social motivations and emotions compete with each other, while higher-level control processes modulate the interactions of these low-level biases.     

The Neural Substrates of Self-Evaluation of Mental Fatigue: A Magnetoencephalography Study

There have been several studies of the neural mechanisms underlying sensation of fatigue. However, little is known about the neural mechanisms underlying self-evaluation of the level of fatigue. The aim of this study was to identify the neural substrates involved in self-evaluation of the level of mental fatigue. We used magnetoencephalography (MEG) with high temporal resolution on 14 healthy participants. During MEG recordings, participants were asked to evaluate their level of mental fatigue in time with execution cues (evaluation trials) or to do nothing in time with execution cues (control trials). The MEG data were analyzed with equivalent current dipole (ECD) and spatial filtering methods to localize the neural activity related to the evaluation of mental fatigue. The daily level of fatigue sensation was assessed using the Checklist Individual Strength questionnaire. In evaluation trials, ECDs were observed in the posterior cingulate cortex (PCC) in seven of 14 participants, with a mean latency of 366.0 ms. The proportion of the participants with ECDs in the PCC was higher in evaluation trials than in control trials (P<0.05, McNemar test). The extent of the decreased delta band power in the PCC (Brodmann’s area 31) 600–700 ms after the onset of the execution cue and that in the dorsolateral prefrontal cortex (DLPFC; Brodmann’s area 9) 800–900 ms after the onset of the execution cue were greater in the evaluation trials than in the control trials. The decrease in delta band power in the DLPFC was positively related to that in the PCC and to the daily level of fatigue sensation. These data suggest that the PCC and DLPFC are involved in the self-evaluation of mental fatigue.     

Spatial learning and action planning in a prefrontal cortical network model

The interplay between hippocampus and prefrontal cortex (PFC) is fundamental to spatial cognition. Complementing hippocampal place coding, prefrontal representations provide more abstract and hierarchically organized memories suitable for decision making. We model a prefrontal network mediating distributed information processing for spatial learning and action planning. Specific connectivity and synaptic adaptation principles shape the recurrent dynamics of the network arranged in cortical minicolumns. We show how the PFC columnar organization is suitable for learning sparse topological-metrical representations from redundant hippocampal inputs. The recurrent nature of the network supports multilevel spatial processing, allowing structural features of the environment to be encoded. An activation diffusion mechanism spreads the neural activity through the column population leading to trajectory planning. The model provides a functional framework for interpreting the activity of PFC neurons recorded during navigation tasks. We illustrate the link from single unit activity to behavioral responses. The results suggest plausible neural mechanisms subserving the cognitive "insight" capability originally attributed to rodents by Tolman & Honzik. Our time course analysis of neural responses shows how the interaction between hippocampus and PFC can yield the encoding of manifold information pertinent to spatial planning, including prospective coding and distance-to-goal correlates.     

The Neural Bases of Directed and Spontaneous Mental State Attributions to Group Agents

In daily life, perceivers often need to predict and interpret the behavior of group agents, such as corporations and governments. Although research has investigated how perceivers reason about individual members of particular groups, less is known about how perceivers reason about group agents themselves. The present studies investigate how perceivers understand group agents by investigating the extent to which understanding the ‘mind’ of the group as a whole shares important properties and processes with understanding the minds of individuals. Experiment 1 demonstrates that perceivers are sometimes willing to attribute a mental state to a group as a whole even when they are not willing to attribute that mental state to any of the individual members of the group, suggesting that perceivers can reason about the beliefs and desires of group agents over and above those of their individual members. Experiment 2 demonstrates that the degree of activation in brain regions associated with attributing mental states to individuals—i.e., brain regions associated with mentalizing or theory-of-mind, including the medial prefrontal cortex (MPFC), temporo-parietal junction (TPJ), and precuneus—does not distinguish individual from group targets, either when reading statements about those targets'' mental states (directed) or when attributing mental states implicitly in order to predict their behavior (spontaneous). Together, these results help to illuminate the processes that support understanding group agents themselves.