The standard definition of mental disorder in clinical psychology and psychiatry is based upon Jerome Wakefield’s ‘Harmful Dysfunction’ model (Wakefield, 1992). According to Wakefield’s theory, for a mental disorder or illness to be present requires a biological mechanism to fail at its naturally selected evolutionary function, whilst simultaneously causing socially defined 'harm'.
While powerful and useful as a general concept, many find this paradigm deeply unsatisfactory. Especially for research and analyses involving neurological mechanisms and their relationship to complex social and interpersonal behaviour. It relies on evolutionary guesswork regarding complex cognition and relies too heavily on abstract, high-level social outcomes. In short: it’s hard to get it to work coherently at multiple levels of abstraction (where in the neurology, cognition, and behaviour of an individual levels of abstraction might be naturally discretised to some extent, rather than imposed arbitrarily as constructs for modeling.)
(There’s a debate in the philosophy of science about the usefulness of ‘levelism’, but in the philosophy of information and in computer science levels of abstraction are regarded as meaningful.)
I propose a framework for modelling mental disorders grounded in objective, frequentist information theory (Shannon, 1948). I argue that psychopathology is best conceptualised not as a vague ‘dysfunction’, but as a highly specific, measurable computational failure: Maladaptive Misinformation Processing (MMP).
A Quick Philosophy of Information Primer
I am writing for psychology researchers, and so a very brief setting of context is prudent. (It’s best to get it out of the way as quickly as possible.)
Psychology researchers should absolutely be suspicious of, and careful with, concepts from philosophy and the philosophy of science. However, when it comes to the nature of information and information processing (and transmission, and encoding, and processing) in nature and in the sciences, the philosophy of information has identified some very real difficulties which do affect conceptualisation (conceptual analysis), operationalisation, and theory.
For the purposes of this short explainer document, the most salient problems are:
1. Semantic Information. The distinction between information in Claude Shannon’s objective-frequentist statistical sense (roughly - the reduction in objective-frequentist statistical uncertainty about the state of a source) and semantic information.
2. The Nature of Information and Conceptual Pluralism: The ongoing argument over the nature of information, and whether pluralism about it should be combined with more rigid conceptions (of which there are many including statistical, logic-based, complexity based, and functionalist.)
3. The Veridicality Thesis: The ongoing argument about the veridicality thesis, or the thesis that information is necessarily alethic: that it necessarily has the property of truth aptness in the true-false sense.
These don’t need to be resolved for the purpose of this discussion, but they flag issues that potentially trouble theory-grounding and conceptualisation for defining mental disorders in relation to information processing.
Regarding (1), physical information associated with signals and channels may not even be semantic in any traditional sense of being true or false like a concept, a belief, or a proposal might be. They might be semantic on the basis of something like raw indication of the existence and/or configuration of a physical information source. Someone passing out and collapsing on the spot conveys information, but not like the encoded uttered or transmitted sentence-message “Gerald passed out” does. The sentence is alethic. The person-passing-out situation, not so much. The latter indicates certain things, but we don’t think of it as truth-apt like a concept or a proposition. Likewise for action potentials.
Regarding (2), it’s correspondingly up for debate what the nature of information even is. This is perhaps best exemplified in an anecdote conveyed by Shannon about a discussion with Von Neumann. Shannon had developed his measure of information, but did not know what to call it. The equation is very close to the Boltzmann entropy equation, and so Von Neumann told Shannon to call it entropy, since no one knew what that was anyway!
Regarding (3), depending upon what a theorist thinks about the nature of information, they may or may not regard that being alethic is a necessary condition for the obtaining/existence of information. Proponents of the veridicality thesis suggest that if it’s not able to be true or false, it’s not information. Opponents reject this. Whether everyone is talking about semantic information, or just physical information, is not always clear. Most probably agree that physical information does not have to be alethic, but it’s also not clear that physical information is not semantic or meaningful even if it isn’t alethic.
That’s enough philosophy of information. It’s probably not good for one’s mental health!
Housekeeping: Noise vs. Misinformation
To understand MMP, and to avoid conceptual confusion in the domain of psychological science, we should first distinguish between standard statistical errors and true misinformation.
Imagine you are in a jungle and hear rustling in the dense foliage. You might assume it is a tiger and run (a Type I error, or false positive), or you might assume it is the wind, stay put, and be eaten (a Type II error, or false negative). Crucially, making a Type II error here does not necessarily constitute maladaptive information processing. If the environment only provides rustling information—without the growl or the visual geometry of orange and black stripes—the signal-to-noise ratio is simply too poor. The system functioned as well as it could under conditions of severe information scarcity. It’s not maladaptive, just starved of adequate information.
Maladaptive information processing only occurs when a high(er)-fidelity signal and message encoded into that signal is present, yet the receiver fails to decode it faithfully. But how does this lead to clinical psychopathology? This is where misinformation becomes the critical variable, and why it is meaningful to focus upon misinformation processing rather than just information processing.
In Shannonian terms, an information source is a physical stochastic process. Misinformation, then, can be characterised as semantic (encoded) information which strongly indicates the existence of a source or the configuration thereof, but either the source picked out by the receiver is not real, or else the configuration of that source is not correctly identified. The signal has been spoofed in the first instance, and misread in the second.
Under the MMP framework, clinical conditions are failures to identify and filter this internal spoofing and/or corruption:
Delusion involves the failure to identify that an internally generated signal (e.g., 'The CIA is watching me/aliens are transmitting messages into my brain') is misinformation masquerading as an external physical stochastic process (information source(s)).
Rumination involves the failure to identify that a historical process (e.g., a past social rejection) is no longer actively transmitting. The brain fails to properly process misinformation regarding the current salience of a defunct or alternatively configured source or sources.
(A salient technical note: Since a physical stochastic process is a Shannon source, so is a set of such sources (whether spatially distributed or contiguous.) Moreover, so is a channel between source and destination.)
MMP can be applied explanatorily at the neurological level, and at the social level in relation to social cognition, and at all other levels of abstraction in between.
Stress-Testing MMP
When introducing a new theoretical architecture to cognitive neuroscience, it must not be simply a restatement of existing paradigms, or else otherwise made redundant by them. During the conceptualisation of MMP, the following candidates became apparent:
1. Predictive Processing
A modern cognitive neuroscientist might argue that MMP is simply Karl Friston’s (2010) Predictive Processing model dressed in different terminology. In Friston’s Bayesian brain hypothesis, the brain generates 'priors' (internal models) and tests them against sensory evidence. Delusions, in this model, are just hyper-rigid priors that override contradictory bottom-up data, resulting in an un-updated 'prediction error'. Is MMP redundant?
2. The Alethic Category Error
In the philosophy of information, theorists like Floridi advocate for the veridicality thesis, whilst others note the Bar-Hillel-Carnap paradox. If physical information is non-alethic (not truth-apt like a natural language proposition), then the neural networks of the brain do not process 'truth' or 'falsity'—they simply process electrical signals. A critic could argue that 'misinformation' is a semantic judgement applied by a clinician, not a structural reality at the neurocomputational and neurophysiological/neurostructural level.
3. Source Monitoring Impairment
Since the 1980s, psychology has utilised the concept of 'Reality Monitoring' or 'Source Monitoring' (Johnson & Raye, 1981) to describe the cognitive process of distinguishing between internally generated imagination and externally derived perception. Is MMP simply a rehash of source monitoring impairment?
4. Isn’t Misinformation Processing Just Bad Information Processing?
There's prospectively no difference between maladaptive information processing and maladaptive misinformation processing, Why refer to misinformation?
Why MMP (Probably) Survives
Far from dismantling the thesis, these objections may just help sharpen MMP. Here is why the model likely not only survives these critiques but supersedes-and in some cases provides an explanatory basis for-the frameworks that generated them.
Counter 1: Shannon over Bayes
The critique of Predictive Processing is foundational. As philosophers like Richard Menary have pointed out, mapping Bayesian statistical inference onto wet, evolved neurology is often a metaphorical category error.
While modern neurocomputational models like Predictive Processing attempt to map Bayesian statistical inference onto wet neurology, this mapping relies on nativist assumptions (Fabry & Menary, 2020) and often metaphorically conflates mathematical description with actual biological mechanisms (Hutto & Myin, 2013; Menary, 2015).
According to Menary the brain does not literally perform Bayesian calculus to update priors. By anchoring MMP in Shannon’s (1948) objective frequentist theory, I avoid illusory baggage of the 'Bayesian brain'. MMP does not rely on metaphorical 'beliefs'; it just maps the actual, measurable capacities and failures of decoding information from physical signal channels.
Counter 2: Indication at the Sub-Personal Level
The alethic paradox can be resolved variously, including by reducing semantic content to indication rather than bivalent truth (Hintikka, 2007). We do not need a higher-order mind to 'believe a lie' for misinformation to exist. Consider hippocampal place cells in a rodent spatial navigation task. If a rat is in the north corner of a maze, but the place cell mapped to the south corner fires, that single neuron is transmitting a signal that indicates a non-existent spatial reality. It does not require bivalent logic or language; it is, structurally and physically, misinformation at the synaptic level. Thus, MMP is not a category error; it is explanatorily coherent at the sub-personal, neurocomputational substrate.
Counter 3: A ‘Fractal’ Explanatory Model
I suggest that this is where the thesis becomes a viable, fundable clinical model.
Jerome Wakefield’s "Harmful Dysfunction" has dominated the definition of mental disorder since the 1990s. But it is notoriously ‘clunky’. Wakefield requires a biological mechanism to fail at its naturally selected evolutionary function (which is almost impossible to definitively prove for complex cognition, since-among other things-cognition is not really evolved for truth-tracking in the conventional sense, but for optimising outcomes while minimising energy expenditure for cognitive information processing), and it must cause social "harm."
Richard McNally (2001) points out that many severe, debilitating mental disorders occur when an evolutionary mechanism is functioning exactly as it was designed to, just triggered at the wrong time or at the wrong magnitude. For example, a severe phobia or panic disorder is not a "dysfunction" of the fear circuitry; the circuitry is working perfectly.
Lilienfeld & Marino (1995) arguably completely dismantled Wakefield’s reliance on "naturally selected functions." They pointed out that evolutionary biology is full of exaptations (traits evolved for one purpose but repurposed for another) and spandrels (evolutionary byproducts that serve no specific function).
I propose that MMP is a superior, scalable replacement. Finally, MMP does not ignore the Source Monitoring framework (Johnson & Raye, 1981); it subsumes it and propbably also explains it. Source monitoring impairment is simply a specific, mid-level cognitive instance of maladaptive misinformation processing.
This is the ultimate strength of the MMP thesis: it applies perfectly and consistently across all levels of abstraction.
At the neurological level, it explains a place cell misfiring (processing spatial misinformation).
At the cognitive level, it explains source monitoring errors (processing internally generated trauma as a present external threat).
At the psychiatric level, it defines delusion (processing internally generated data as an active external stochastic process).
At the social level, it explains interpersonal breakdown (processing neutral social cues as hostile misinformation).
Counter 4: We have the word ‘Misinformation’ for a reason.
Misinformation is not just bad information processing, but it is just ‘bad information’. Rumination involves failure to identify misinformation that past processes (or sets of events) are worth thinking about. Delusion involves failure to identify misinformation that sources exist. Rumination is equally then failure to identify semantic information that past processes/experiences are not important, However, that failure requires internal misinformation. Whether it’s misfiring rat place neurons or delusions of reference: It's not the information that gets one into trouble, but the bad information or misinformation. Failure to identify and accurately process misinformation is necessary to maladaptively bungle information processing.
Conclusion
Functionalism is a powerful explanatory paradigm, but perhaps we do not need to rely on the vague evolutionary guesswork of the 'Harmful Dysfunction' to define mental illness? Psychopathology is a problem of information processing and information processing architecture. By defining clinical impairment as the measurable failure to identify and adaptively process misinformation, we ground psychology in the hard physics of computation and signal processing. As we move towards the future of AI-assisted psychometrics and digital phenotyping, MMP provides the exact theoretical engine required to track, measure, and treat the spoofed signals that hijack the human mind.
References
Folstein, J. R., & Van Petten, C. (2008). Influence of cognitive control and mismatch on the N2 component of the ERP: A review. Psychophysiology, 45(1), 152–170.
Fabry, R. E., & Menary, R. (2020). The enculturated predictive processing framework. In Smortchkova, J., Dołęga, K., & Schlicht, T. (Eds.), What are mental representations? (pp. 206–233). Oxford University Press.
Friston, K. (2010). The free-energy principle: A unified brain theory? Nature Reviews Neuroscience, 11(2), 127–138.
Hintikka, J. (2007). Socratic epistemology: Explorations of knowledge-seeking by questioning. Cambridge University Press.
Johnson, M. K., & Raye, C. L. (1981). Reality monitoring. Psychological Review, 88(1), 67–85.
Lilienfeld, S. O., & Marino, L. (1995). Mental disorder as a Roschian concept: A critique of Wakefield's "harmful dysfunction" analysis. Journal of Abnormal Psychology, 104(3), 411–420. https://doi.org/10.1037/0021-843X.104.3.411
McNally, R. J. (2001). On Wakefield's harmful dysfunction analysis of mental disorder. Behaviour Research and Therapy, 39(3), 309–314. https://doi.org/10.1016/S0005-7967(00)00069-4
Menary, R. (2015). The enculturated brain. In T. Metzinger & J. M. Windt (Eds.), Open MIND (pp. 1–28). MIND Group. https://doi.org/10.15502/9783958570085
Shannon, C. E. (1948). A mathematical theory of communication. The Bell System Technical Journal, 27(3), 379–423.
Wakefield, J. C. (1992). The concept of mental disorder: On the boundary between biological facts and social values. American Psychologist, 47(3), 373–388.
