The relationship between subpersonal priors from active inference theory and parts from the Internal Family Systems (IFS) model reveals fascinating parallels between cognitive neuroscience and psychotherapy. Though emerging from different disciplines—active inference from computational neuroscience and IFS from psychotherapy—these frameworks offer complementary perspectives on how the mind processes information, manages internal conflicts, and responds to the environment.
Foundational Concepts and Structural Parallels
Multiplicity of Mental Processes
Both frameworks fundamentally reject the notion of a unitary mind. IFS, developed by Richard Schwartz in the 1980s, “recognizes the mind as a naturally subdivided entity, capable of supporting many parts or sub-personalities”1. Similarly, active inference views cognition as involving multiple competing predictive models or priors that operate below conscious awareness.
This multiplicity serves crucial functions. In IFS, “the non-extreme intention of each part is something positive for the individual”4, while in active inference, different subpersonal priors help the brain resolve ambiguity and optimize behavior across varied contexts.
Hierarchical Organization
Both frameworks employ hierarchical structures. IFS positions the Self as the central coordinator designed to lead the internal system, with various parts (Managers, Firefighters, and Exiles) fulfilling specialized roles. The goal of IFS therapy is to “differentiate and elevate the Self so it can be an effective leader in the system”4.
Active inference similarly describes hierarchical processing where “higher hierarchical levels regulate lower levels by setting their preferred or predicted outcomes (or set points), which lower levels realize”25. This allows for higher-order contextual modulation of lower-level processes.
Mapping Specific Components
Managers as Control Priors
IFS Managers serve a protective, regulatory function—they “maintain control and protect the individual from pain”17. These parts develop to maintain stability and avoid distress.
In active inference, this maps to control processes that propagate goals through structured plans or policies. The “control hierarchy propagates goals through structured plans or policies”25, similar to how Managers implement protective strategies. Just as Managers try to maintain system stability, control-oriented priors in active inference aim to minimize prediction errors through planned action sequences.
Exiles as Encapsulated Prediction Errors
IFS Exiles represent “young parts that develop because of a traumatic experience”16, carrying painful emotions that have been separated from conscious awareness. They “often isolate themselves from the individual to protect them from pain and fear”16.
These map to what might be considered encapsulated prediction errors in active inference—sensory experiences that couldn’t be adequately predicted or integrated. Just as “some ‘unusual’ signals—those that violate predictions—receive special attention and are often interpreted as ‘bad feelings, negative emotions, or pain’”29, Exiles carry emotional pain that couldn’t be processed through typical predictive mechanisms.
Firefighters as Emergency Response Priors
IFS Firefighters “emerge when Exiles break out and demand attention”16, creating rapid, sometimes maladaptive responses to suppress emotional pain. Their defensive reactions can be impulsive and extreme.
In active inference terms, these might represent emergency response priors that activate when prediction errors suddenly increase beyond manageable levels. These priors prioritize immediate distress reduction over long-term optimization, similar to the “reactive processes [that] quickly define subjective experience, allowing little room for any testing of these potentially distorted beliefs against reality”28.
Self as Meta-Cognitive Integration
The IFS Self represents “a resourceful, calm, and intact whole within”5 that ideally coordinates the parts system. When functioning optimally, it processes information holistically and responds adaptively.
This parallels meta-cognitive processes in active inference that integrate information across multiple generative models. As noted in research on multiple internal models, “one could imagine that they underwrite some conscious inference, with several competing generative models (i.e., hypotheses) running at a subpersonal or unconscious level in the brain”32.
Functional Similarities
Precision and Confidence
IFS therapy aims to help clients differentiate parts from the Self, recognizing when parts are activated and their degree of influence. This parallels precision weighting in active inference, which “determines the relative influence of control (priors) versus motivation (sensory evidence)”25.
High-precision priors strongly influence perception and behavior, just as highly activated parts can dominate the internal system. As search result 27 notes, “precision can be conceptualized as the inverse of uncertainty; highly precise signals are weighted more heavily in perceptual inference”27.
Integration vs. Dissociation
Both frameworks address issues of integration versus dissociation. IFS therapy aims for “balance and harmony within the internal system”4, helping parts release their burdens and work cooperatively.
Active inference similarly addresses the integration of competing models and priors. When integration fails, mental disorders may emerge: “certain psychiatric disorders, especially those characterized by chronic and unrelenting anxiety, are preferentially susceptible to top-down constructed dysfunctions”28. This parallels the IFS view that psychological distress emerges when parts become extreme in their roles.
Practical Mapping and Therapeutic Implications
Parts Mapping and Model Selection
The IFS parts mapping process “serves to identify the Parts in our individual system”3 and their relationships, allowing clients to recognize which parts are active in different situations.
This parallels model selection in active inference, where the brain must determine which internal model best explains current sensory data. As result 32 describes, “when receiving sensory input generated by a particular conspecific, how does an animal know which internal model to update?”32 This is analogous to the IFS question of identifying which part is currently activated.
Therapeutic Change Processes
Both frameworks offer pathways for change through similar mechanisms:
- Awareness and identification: IFS therapy begins with identifying parts, while active inference requires identifying which priors are generating predictions.
- Unburdening/updating: IFS involves “unburdening” parts of their negative beliefs, while active inference involves updating maladaptive priors.
- Integration: IFS aims for “harmony within a person’s internal system”1, while active inference seeks optimal integration of predictive models.
Conclusion
The mapping between subpersonal priors in active inference and parts in IFS reveals striking conceptual parallels between these frameworks. While originating from different disciplines and using different terminology, both offer complementary perspectives on the multiplicity of mind, hierarchical organization, and processes of integration.
This mapping suggests that psychological healing might involve similar mechanisms whether conceptualized as updating maladaptive priors or unburdening parts carrying emotional pain. Understanding these parallels may help bridge the gap between neuroscientific and psychotherapeutic approaches to mental health, offering multiple conceptual tools to understand the complexity of human experience.