Active inference presents a revolutionary framework for understanding how attitudes shape human cognition and behavior through the integrated functioning of motivational and cognitive brain systems. This integration occurs primarily through the ventromedial prefrontal cortex (vmPFC) and dorsolateral prefrontal cortex (dlPFC), which together form a neural substrate for attitudes as evaluative frameworks that guide perception, decision-making, and behavior.
The Traditional Division of Cognitive and Motivational Processes
Historically, neuroscience has maintained a functional division between cognitive and motivational processes in the prefrontal cortex. The dlPFC has been primarily associated with cognitive control and executive functioning, while the vmPFC has been linked to emotional processing and value-based decision-making. This functional segregation is reflected in the common understanding that “the vmPFC is assumed to have a crucial role in emotional processing, whereas the dlPFC is predominantly involved in cognitive control and executive processing”9.
This traditional perspective defines the vmPFC as “sensitive to the reward or value of stimuli, value-based decision-making, anticipation of reward and self-based evaluation,” while the dlPFC is involved in “working memory, divergent thinking, executive attention, and decision making”9. However, this strict functional distinction has been increasingly challenged by evidence showing that both regions contribute to both cognitive and emotional processing, albeit in different ways.
Active Inference: A Unifying Framework
Active inference provides a powerful theoretical framework for understanding how attitudes, motivation, and cognition interact. It views the brain as “a statistical organ that forms internal generative models of the (hidden) states and contingencies in the world, and uses these models to continuously generate predictions in the service of perception and adaptive behavior”24.
The framework proposes that all brain function, including perception, learning, and action, serves a single imperative: to minimize free energy or “surprisal.” This occurs through two complementary processes:
- Perception minimizes free energy “by (Bayesian) belief updating or changing your mind, thus making your beliefs compatible with sensory observations”
- Action minimizes free energy “by changing the world to make it more compatible with your beliefs and goals”8
This unification of perceptual and motor functions under a single principle offers a conceptual bridge between traditionally separated cognitive and motivational domains.
Attitudes as Prior Beliefs in Active Inference
Within the active inference framework, attitudes can be conceptualized as prior beliefs that shape how we perceive and interact with the world. In active inference, “perception is not what we sense but a computational compromise between our expectation of what we believe we should be sensing and the actual sensation experienced”9. Similarly, attitudes function as evaluative frameworks that guide our expectations and interpretations of experiences.
Gordon Allport’s definition of an attitude as “a mental and neural state of readiness, organized through experience, exerting a directive or dynamic influence upon an individual’s response” aligns perfectly with the active inference view of prior beliefs. Both represent expectation-laden frameworks that shape perception and guide action.
Active inference proposes that “believing has recently been recognized as a fundamental brain function linking a person’s experience with his or her attitude, actions and predictions.” This process leads to “probabilistic neural representations that allow individuals to develop preferences, maintain positions, and to tailor behavior”15.
The vmPFC: Prioritizing Goals Through Valuation
The vmPFC plays a crucial role in integrating motivation into the active inference framework. Research has shown that the vmPFC is involved in processing “motivational variables – such as incentives and expected values – and metacognitive variables – such as confidence judgments”3.
In active inference terms, the vmPFC helps to prioritize goals by assigning value to different possible outcomes. This valuation process allows the brain to determine which goals should receive greater precision (or confidence), thereby influencing action selection. As noted in the research, “the incentive value of an outcome corresponds to its prior (log) probability, so that preferred outcomes (or goals) have high prior probability”24.
The vmPFC is particularly important for self-relevant processing, with studies showing “greater activation in the vMPFC for objects assigned to the self compared to objects assigned to the other person”12. This self-relevance processing is crucial for determining which goals are most important for the organism.
The dlPFC: Propagating Goals Through Cognitive Control
While the vmPFC prioritizes goals through valuation, the dlPFC enables goal-directed behavior through cognitive control. Research indicates that “one well-documented role of the dlPFC is maintenance and regulation of top-down control for driving appropriate behavior”79.
In active inference, the dlPFC helps to propagate goals through hierarchical levels of processing. This propagation allows abstract goals to influence concrete behaviors through what active inference theorists call “deep goal hierarchies.”
The dlPFC appears to be particularly important for controlling emotional responses to align with goals. Studies show that “anodal tDCS over the dlPFC altered valence attribution to emotional pictures,” suggesting that the dlPFC plays a role in regulating emotional responses to support goal-directed behavior79.
Neural Integration of Motivation and Cognition
The integration of motivational and cognitive processes occurs through the interaction between the vmPFC and dlPFC. Research using Dynamic Causal Modeling (DCM) has revealed that “the effective connectivity from left dlPFC-BA46 to vmPFC plays a critical role in self-control”1. This connectivity allows cognitive control mechanisms in the dlPFC to modulate value computations in the vmPFC.
Studies show that during self-control tasks, the dlPFC “comes online and modulates activity in vmPFC so that its value computations incorporate” abstract attributes rather than just immediate, concrete ones1. This modulation is crucial for behavior that prioritizes long-term goals over immediate rewards.
The active inference framework characterizes this integration by proposing that “control and motivation (implemented mainly in dorsal and ventral neural streams, respectively) conspire to propagate and prioritize goals, respectively, in the service of goal-directed action”24. This integration “results in a joint optimization of action sequences (and state transitions) and their precision”24.
Hierarchical Processing: From Attitudes to Behavior
Active inference emphasizes that cognition and motivation operate through hierarchical processing. At higher levels of this hierarchy, abstract attitudes and goals are represented, while lower levels translate these into concrete actions.
This hierarchical structure allows attitudes to influence behavior through a cascade of prior beliefs propagating down the hierarchy. As higher-level attitudes constrain lower-level expectations, they shape perception and guide action selection.
Research suggests that “the control hierarchy propagates prior preferences or goals, but their precision is informed by the motivational context, inferred at different levels of the motivational hierarchy”11. This creates a bidirectional flow where attitudes shape perception and action, while experiences update attitudes through prediction errors.
The Filtering Function of Attitudes
One crucial aspect of attitudes in active inference is their filtering function. Attitudes, implemented through the interaction of the vmPFC and dlPFC, determine which aspects of sensory information reach consciousness and how they are interpreted.
The dlPFC acts as a cognitive filter, selecting which information is relevant for current goals. Studies show it “is involved in control and regulation of valence of emotional experiences”79. Meanwhile, the vmPFC operates as a motivational filter, determining the significance or value of information. Research indicates it “might be involved in the extinction of arousal caused by emotional stimuli”79.
Together, these filtering mechanisms create what active inference theorists call “precision-weighted prediction errors” – signals that determine how strongly sensory information influences beliefs and behavior. High-precision attitudes are less susceptible to updating by contradictory evidence, explaining phenomena like confirmation bias.
Clinical Implications and Disorders of Integration
The integration of attitudes, motivation, and cognition through vmPFC-dlPFC interaction has important clinical implications. Dysfunction in this system can lead to various psychological disorders:
“Hyperactivity of the vmPFC, and the amygdala due to insufficient control via the dlPFC leads to an attention bias to threat-related stimuli in anxiety”79. Similarly, “a lack of regulatory control of the dlPFC on dysregulated fear circuits in the vmPFC is involved in post-traumatic stress disorder (PTSD)”79.
Depression may involve “reduced dlPFC control over the vmPFC or an imbalance of respective interactions”79. From an active inference perspective, these conditions reflect dysfunctional precision-weighting, where negative attitudes receive excessive precision while positive ones are underweighted.
Treatment approaches targeting the integration of motivation and cognition show promise. For instance, “increasing the excitability of dlPFC led to modulation of interpretation bias in valence attribution to the positive stimuli”7, suggesting a potential mechanism for therapeutic interventions.
Conclusion: An Integrated Model of Attitudes in Active Inference
Active inference provides a compelling framework for understanding how attitudes integrate motivational and cognitive processes through the interaction of the vmPFC and dlPFC. Attitudes function as prior beliefs that shape perception, valuation, and action selection through hierarchical processing in these brain regions.
The vmPFC contributes to this process by prioritizing goals through valuation, while the dlPFC propagates goals through cognitive control. Their integration allows attitudes to guide behavior in a context-sensitive manner that balances immediate rewards with long-term objectives.
This integrated perspective has implications beyond basic neuroscience, informing our understanding of psychological disorders, decision-making processes, and potential therapeutic interventions. By conceptualizing attitudes within the active inference framework, we gain insight into how these evaluative frameworks shape our subjective experience and guide our interactions with the world.