Behaviourism, Identity Theory and Functionalism

Problem Chapter 76 brought the ontological and causal questions of the mind–body relation to the fore. The question now is: how are mental states to be identified philosophically? When we speak of “pain,” “belief,” “desire,” “memory,” “fear,” or “decision,” are we speaking of a pattern of behaviour, a brain state, or a causal role that may be realised in more than one physical system? In twentieth-century philosophy of mind, behaviourism, mind–brain identity theory, and functionalism emerged as three important answers. All three share an anti-Cartesian tendency, but they are not one and the same theory. Behaviourism attempts to understand mental talk in the language of behaviour and dispositions; identity theory treats a mental state or process as a brain state or process; functionalism takes its causal-functional role to be decisive. The term “behaviourism” is itself multivalent. Methodological behaviourism emphasises a research method in psychology centred on publicly observable variables, stimulus, response, and learning history. Psychological or radical behaviourism gives a larger role to environmental contingencies in explaining behaviour. Logical or philosophical behaviourism seeks to analyse the meaning of mental concepts, or the conditions for attributing them, through behavioural dispositions. Without these distinctions, Ryle, Skinner, and laboratory behaviourism are easily and misleadingly treated as one thing. The attraction of logical behaviourism lies in verification, public criteria, and distance from the Cartesian “private theatre.” If saying “Sita is in pain” means, at least in part, that she is disposed to groan, protect the injured part, seek treatment, or report pain, mental vocabulary remains connected with the empirical world. But the difficulty appears immediately. A person may be in pain without groaning; an actor may groan without being in pain; a stoic may suppress expression; paralysis may prevent behaviour altogether. The relation between a mental state and overt behaviour may therefore be regular without being an identity. Behavioural dispositions also depend on background beliefs and desires. Believing that it is raining does not by itself determine the behaviour of taking an umbrella: whether one wants an umbrella, intends to go out, minds getting wet, or trusts a weather app all involve other mental states. This holism makes a pure behaviouristic reduction difficult. Identity theory offers a different response to this difficulty. Sensations, experiences, and mental processes are not merely shorthand for behaviour; they are real internal states. But they are not states of a non-physical substance: they are brain processes. The mid-twentieth-century formulations of U. T. Place, J. J. C. Smart, and Herbert Feigl were decisive in developing this line of thought. A classic model for identity theory is the scientific identity “lightning = electrical discharge.” “Pain” and “C-fibre firing” need not mean the same thing; empirical discovery may show two different descriptions to concern one underlying phenomenon. The pain–C-fibre example is historical and illustrative, however; contemporary neuroscience does not claim so simple a one-to-one mapping. Here type identity and token identity must be distinguished. Type identity says that a mental kind, such as pain, is identical with a physical kind. Token identity makes a weaker claim: each particular mental event is some particular physical event, even if the same mental type is realised differently in different organisms or contexts. The objection from multiple realisability puts pressure on type identity. If a human being, an octopus, a hypothetical silicon system, or organisms with different neural architectures could all instantiate the same functional pain-state, identifying pain with one specific human neural type appears too narrow. This was an important historical motivation for the functionalism associated with Putnam and Fodor. Functionalism holds that the identity of a mental state is determined primarily not by its internal material composition but by its causal role within a system: what inputs produce it, how it relates to other internal states, and what tendencies towards behaviour or output it generates. For example, a pain-state may be identified by a causal-role pattern in which damage or other input produces it, it interacts with beliefs and desires associated with distress, and it tends to generate avoidance, reporting, or care-seeking. In principle, this role may be realised by a biological brain, by another physical substrate, or by a different architecture. Functionalism preserves behaviourism’s concern with public criteria while accepting the reality of hidden internal causal states. It is therefore tempting to call it a bridge between behaviourism and identity theory. Yet functionalism itself divides into several forms: machine-state functionalism, causal-theoretical functionalism, psychofunctionalism, and analytic functionalism. The most difficult question concerns phenomenal consciousness. If two systems have the same functional organisation but differ, or perhaps one lacks altogether, the inner experience of “seeing red,” is functional role sufficient? Thought experiments involving inverted qualia, absent qualia, and philosophical zombies expose this tension. Chapter 78 will examine the problem of consciousness independently and in greater depth. Intentional content is another challenge. A belief is not merely an input-output role; a belief is about something. Does functional role determine content, or are the external world, history, language, and social practice also constitutive? Chapter 79 will take up this question in detail. The age of AI gives this old debate a new form. If an LLM or a robot produces human-like answers or behaviour, should we attribute mental states to it? Behaviourist criteria may be relatively permissive; functionalism will examine architecture and role; a biologically brain-specific identity theory may hesitate. But behaviour alone, functional organisation, and consciousness are not the same question. The problem of this chapter is therefore not simply, “Which of the three theories is correct?” Its aim is to clarify the relations among criteria for mental-state attribution, the underlying realiser, causal role, multiple realisation, levels of explanation, subjective experience, and empirical evidence. Methodological rule: philosophical behaviourism is not Skinnerian psychology; identity is not correlation; functionalism is not merely a “computer metaphor”; multiple realisability is not a denial of physicalism; and fluent AI output is not evidence of consciousness. Without these limits the whole debate turns into caricature. Core Propositions The core proposition is that behaviourism, identity theory, and functionalism should not be read as an evolutionary ladder—“the first was wrong, the second incomplete, the third final.” Each illuminates a different dimension of mental explanation, and some insight from each survives in transformed form within later theories. The enduring insight of behaviourism is that mental concepts, in their public use, are not wholly separate private objects detached from action, disposition, learning, expression, and interpersonal criteria. The semantic and pragmatic force of mental attribution is deeply related to behaviour. Yet treating a mental state as an exhaustive translation into actual or hypothetical behaviour fails because behaviour is context- sensitive, simulation is possible, suppression is possible, disability may alter output, and mental states stand in complex causal relations to one another. The enduring insight of identity theory is that the internal physical realisation of mental life is a serious fact. Experience is not merely public behaviour: neural injury, pharmacology, stimulation, anaesthesia, and disease alter mental states. Brain-state causation is not a peripheral detail in a theory of mind. गजेन्द्र ठाकु र But a claim of scientific identity requires empirical detail. The broad thesis “pain is a brain process” differs from the specific type-identity claim “pain is exactly neural type X.” Contemporary evidence concerning variability, distributed processing, and plasticity does not support an overly simple type-level identity. The enduring insight of functionalism is that organisation and causal role matter. By individuating a state through its relations to other states, inputs, and outputs within a system, one preserves the explanatory generality of mental kinds. Multiple realisation is a powerful motivation for functionalism: the same function may be fulfilled by different physical structures. But “could be realised differently” is not equivalent to “actual human mental kinds are independent of the brain.” The biological constraints on human cognition remain empirically important. Functionalism is compatible with physicalism, but it is not logically identical with identity theory. A functional state may be a higher-level role instantiated by a physical realiser. Hence “functionalism = non-physicalism” and “functionalism = computer dualism” are both mistaken. The distinction between role and realiser is decisive. A thermostat state may have a functional role within a temperature-control circuit, while its realiser is particular hardware. In the theory of mind, the question is whether by “pain” we mean the role or the neural realiser; the answer may vary with context. Token identity can coexist with functionalism: a particular functional mental event may be a particular physical event while the same functional type can be realised in multiple physical types. The boundaries among the historical schools are therefore porous. A causal-role account is attractive for the problem of mental causation. Belief–desire states produce behaviour because their functional identities lie within a causal network. Yet if the underlying neural mechanism is the real physical cause, whether role- level causation is genuine or derivative remains an open question. The qualia objection limits claims of functional sufficiency. If functional duplicates could differ in phenomenal experience, functional organisation would not by itself exhaust consciousness. Even if such cases are impossible, their possibility has to be argued rather than assumed. The question of intentionality requires functionalism to have a theory of content. Internal causal role may be one component of broad content, but the contribution of reference, history, environment, and social norms cannot simply be ignored. Behavioural evidence, neural evidence, and functional or computational evidence are three distinct streams of evidence. A reliable philosophy of mind requires a multi-evidential method in which one evidence stream does not automatically settle the questions raised by another. In clinical diagnosis, behaviour matters, but so do subjective report, physiological evidence, developmental history, and context. In neuroscience, brain data matter, but interpretation is difficult without task design and behaviour. In AI, output matters, but architecture, training, persistence, embodiment, agency, and valence remain separate questions. The parallel conclusion of this chapter is: “Behaviour is the public doorway to mental attribution; the brain is the principal basis of its biological realisation; functional organisation is its explanatory pattern. Examine the relations among all three levels, and do not turn any one of them into the whole mind.” Principal Arguments First argument—the Rylean critique of the category mistake. The Cartesian picture turns mind into a “ghostly thing” parallel to the body. Ryle’s important point is that explaining intelligent action every time by a hidden inner episode risks either an infinite regress or a category error. Skill is displayed in organised dispositions to act. Calling Ryle a strict reductive behaviourist is controversial; his project was to clarify the logical geography of ordinary mental concepts. This chapter therefore does not reduce Ryle’s insight to the slogan “the mental = bodily movement.” Second argument—the language of dispositions. To say that glass is fragile does not mean that it is now broken, but that under suitable conditions it is disposed to break. Similarly, attribution of belief or desire may be connected with patterns of counterfactual behaviour. Yet specifying the antecedents and consequents of a mental disposition may itself require other mental terms, producing circularity. Third argument—the super-Spartan/actor objection. A person may suppress a report of pain because of training, fear, or paralysis, while an actor can display convincing pain behaviour without pain. Observable behaviour is therefore neither a necessary nor a sufficient identity condition for pain. Fourth argument—the Place–Smart model of scientific identity. Even if “sensation” and “brain process” are not synonyms, an empirical identity may still hold. The analogy of lightning and electrical discharge shows that semantic difference and ontological identity can coexist. Fifth argument—Occam and nomological economy. If the systematic physical correlates of mental events and their causal work can be explained through brain processes, adding independent psychophysical “danglers” may violate theoretical economy. Feigl’s physicalist motivation can be read in this direction. Sixth argument—the predictive promise of type identity. If a particular mental kind could be linked to a stable neural kind, neuroscientific explanation would become more unified. But actual neural heterogeneity, developmental plasticity, species variation, and network-level coding may obstruct any simple one-to-one mapping. Seventh argument—the Kripkean modal objection. If “pain = neural state” is an identity, and identities are necessary, how can we conceive pain without that neural state or that neural state without pain? An identity theorist may reply that phenomenal concepts, epistemic access, and metaphysical necessity must be distinguished. This debate is kept here only up to the point where it touches the qualia problem that Chapter 78 will address. Eighth argument—multiple realisability. One motivation in Putnam is that, if psychological states are defined as machine states or functional organisations, different physical species or systems may realise the same mental state. This increases pressure on a strict, species-independent type identity. Ninth argument—functional definition. A mental state M can be identified within a causal network: input I produces M; M together with belief or desire N tends to produce output O; M also modifies other states. This is richer than a purely behaviouristic I–O mapping because it explicitly includes internal causal relations. Tenth argument—the Ramsey/Lewis style of theoretical identification. A folk-psychological theory may specify mental states by their causal roles; empirical science may later identify the physical states that uniquely realise those roles. Functional role and physical realiser can therefore be consecutive explanatory stages rather than competitors. Eleventh argument—machine functionalism. A Turing-machine-style description specifies state-transition organisation rather than physical material. This abstraction is powerful in computational cognitive science. Its limitation is that human cognition also involves continuous, embodied, affective, and socially embedded processes. Twelfth argument—psychofunctionalism. Internal information-processing roles specified by empirical cognitive psychology may provide the basis for a taxonomy of mental states. This differs from analytic or common-sense functionalism because the theory of a mature science, rather than ordinary concepts alone, may be decisive. Thirteenth argument—the Block/Fodor objection to simple functional analysis. Distinct psychological states may generate the same coarse-grained input-output behaviour, and the same state may produce different behaviour in different contexts. Internal architecture and relations are therefore necessary. Fourteenth argument—inverted qualia. Suppose two systems are functionally identical, yet their experiences of red and green are reversed. If this is coherent, functional description underdetermines phenomenal character. The functionalist may deny the गजेन्द्र ठाकु रक समानान्तर दर्शन — खण्ड २ possibility, enrich the account with higher-order or representational features, or restrict functionalism to non-phenomenal states. Fifteenth argument—absent qualia and the China-brain intuition. Imagine a vast distributed system that imitates human functional organisation but seems to lack “experience.” This intuition challenges functional sufficiency. Yet intuition is not itself the final evidence; the details of implementation, integration, and counterfactual structure are decisive. Sixteenth argument—causal efficacy. Behaviourism risks dissolving mental causes into behavioural regularities; identity theory gives mental causes a causal place by identifying them with neural causes; functionalism seeks to preserve causation at the level of causal roles. The three offer different strategies for the problem of mental causation. Seventeenth argument—explanatory pluralism. “He ran because he was afraid,” “his salience or amygdala circuitry was active,” and “a threat-avoidance control state was activated” may answer different questions about the same event. Ontological economy does not require a single explanatory vocabulary. Eighteenth argument—the AI test. Turing-style behavioural success may be important evidence for attributing intelligence, while functionalism may demand the relevant counterfactual causal organisation of the architecture. Surface imitation of text is not full functional equivalence. Consciousness is a third and separate question. Nineteenth argument—embodiment. Classical functionalism often adopts an abstract input–internal-state–output model; embodied and enactive critiques argue that the body–environment loop may itself be constitutive of cognition. Functionalism can accommodate this challenge by widening what counts as the causal role. Twentieth argument—normativity. A belief is not merely a causal state; it can be true or false, rational or irrational. A purely causal-role theory must explain the source of normativity—social practice, teleology, evolutionary function, inferential role, external relations, or some combination. Twenty-first argument—discipline about levels. Psychological kinds need not be coextensive with neurological kinds, but higher- level sciences are not thereby illegitimate. Economics, genetics, and ecology all form categories at multiple levels; the meaning of reduction in psychology must likewise be specified carefully. Twenty-second argument—underdetermination by evidence. The same behavioural dataset may receive different interpretations within behaviourist, identity-theoretic, and functionalist frameworks. Theory choice must therefore consider multiple criteria: explanatory scope, prediction, fit with neuroscience, phenomenology, parsimony, and resistance to counterexamples. Pūrvapakṣa Pūrvapakṣa 1: “Mental states are not directly observable; scientific practice should therefore admit only observable behaviour.” The challenge to this argument is that science also accepts unobservable entities—electrons, genes, black holes, latent variables —on the basis of indirect evidence. Unobservability alone is not a reason for elimination. Pūrvapakṣa 2: “Behaviour is sufficient; hypotheses about inner states are metaphysical luxuries.” But the same behaviour may have different internal causes, and the same internal state may produce different behaviour. Learning, deception, suppression, paralysis, and context make this gap real. Pūrvapakṣa 3: “Ryle proved that inner mental states are a category mistake.” This is an excessive simplification of Ryle. The category-mistake critique targets a Cartesian ghostly substance; the claim that Ryle denied the reality of all sensations and episodes requires much greater caution. Pūrvapakṣa 4: “As neuroscience advances, mental vocabulary will disappear.” History shows that higher-level categories often become more precise after mechanisms are discovered rather than disappearing. Terms such as memory, attention, and language are refined by neural evidence. Pūrvapakṣa 5: “If pain is a brain process, we ought to be able to see pain through a microscope.” Identity theory distinguishes semantic or perspectival access: water can be H2O without thirst feeling like a chemical formula. Identity does not abolish the mode in which an experience is presented. Pūrvapakṣa 6: “The meaning of the word ‘pain’ differs from the meaning of brain terms; therefore the identity is false.” Co- reference and identity do not depend on synonymy. “Morning Star” and “Evening Star” can differ in cognitive significance while referring to Venus, although the mental case introduces additional phenomenal issues. Pūrvapakṣa 7: “Different species have different brain patterns for pain; therefore physicalism is false.” Multiple realisability attacks some versions of type identity, not physicalism as such. Every realisation could still be physical. Pūrvapakṣa 8: “Functionalism proves that the mind is merely software.” Functionalism may individuate mental states by role, but this does not make the software/hardware analogy a universal ontology. Biological dynamics, embodiment, and affect can be included within the functional specification. Pūrvapakṣa 9: “Same function implies same experience.” This is among functionalism’s most controversial phenomenal claims. Inverted and absent qualia objections challenge the implication; it should not be treated as an established fact. Pūrvapakṣa 10: “If a computer produces the same input-output pattern, its mind is human-like.” Input-output parity may meet some logical-behaviourist criteria, but internal causal architecture, embodiment, semantic grounding, persistence, self-model, affect or valence, and consciousness remain separate empirical and philosophical questions. Pūrvapakṣa 11: “The Chinese Room decisively refutes functionalism.” Searle’s argument gives a powerful intuition concerning the syntax–semantics gap, but systems replies, robot replies, expanded causal-role accounts, and other responses keep the debate open. It should not be presented as a final knock-down proof. Pūrvapakṣa 12: “Because functionalism accepts multiple realisability, brain research is irrelevant.” On the contrary, neuroscience is indispensable for understanding actual human realisation. Functional abstraction may be a model built upon biological constraints, not a replacement for them. Pūrvapakṣa 13: “Identity theory is reductionist, therefore person-level reasons are fictitious.” Ontological identity does not entail explanatory elimination. The realisation of temperature by molecular motion does not make thermodynamics meaningless, though the limits of the analogy with mind must be tested separately. Pūrvapakṣa 14: “Behavioural criteria can be oppressive, therefore behaviourism is wholly useless.” Observable behaviour is valuable data in clinical and educational assessment. The problem arises when it is made the sole truth in place of subjective report, context, disability, or covert processes. Pūrvapakṣa 15: “Indian philosophy had already stated functionalism.” Some Indian theories contain subtle accounts of the causal role of cognition, the body–sense–mind relation, or inference from signs. But the computational, machine-state, and multiple-realisability genealogy of modern functionalism is different. Functional similarity is not historical identity. Pūrvapakṣa 16: “Ātman = Cartesian mind, manas = brain.” This taxonomy is false. Nyāya’s relations among ātman, manas, the senses, body, and cognition as a quality do not map one-to-one either onto Cartesian substance dualism or contemporary neuroscientific categories. Pūrvapakṣa 17: “Behaviourism strengthens moral responsibility because only action should be considered.” Legal and ethical evaluation also asks about intention, knowledge, capacity, coercion, reasons, and context. Observable action is necessary evidence in many cases, but it is not a complete mental attribution. Pūrvapakṣa 18: “Functionalism is value-neutral.” The selection of inputs and outputs, assumptions about normal behaviour, and choices of training data can carry social norms into functional categories. Normativity can be hidden inside mental-state technologies. गजेन्द्र ठाकु र Uttarapakṣa The first rule of the uttarapakṣa is to read the three theories as answers to different questions. Behavioural evidence illuminates “How do we attribute mental states?” Identity theory asks, “What is the physical realisation of a human mental event?” Functionalism asks, “What is the causal organisation of a mental kind?” A revised behaviourist lesson does not reduce mental life to behaviour; it treats behaviour as evidential and as a public criterion. Proper inquiry triangulates first-person report, physiological measures, task performance, longitudinal patterns, and social context. Ryle’s lasting lesson is not the elimination of an inner ghost but discipline about categories: knowing how, intelligent performance, disposition, competence, and action should not be explained by introducing an unnecessary hidden duplicate. A strong form of identity theory can accept broad token physical realisation while leaving strict one-type/one-neural-type claims to empirical evidence. This preserves fit with neuroscience while acknowledging the pressure of multiple realisability. Keep correlation distinct from identity. An fMRI correlate may be an indicator of a mental state, not its identity. An identity thesis requires causal, theoretical, cross-contextual, and ontological argument; an imaging “blob” alone is not enough. Functionalism’s role–realiser distinction can illuminate the division of labour between psychology and neuroscience. Psychological theory specifies causal roles; neuroscience investigates actual human realisers or implementations; computational models explore transition structures. Treat multiple realisability as a graduated hypothesis. Some mental functions may be highly substrate-flexible; some affective or interoceptive states may depend more heavily on specific biological organisation. Empirical science should determine the degree. Functional equivalence is not superficial similarity of output. Counterfactual robustness, learning, error correction, interaction among internal states, temporal persistence, context sensitivity, and responses to perturbation should be tested. For AI, a benchmark score is therefore not enough. A system may give the same answer through memorisation, a shortcut, retrieval, simulation, or a robust internal model. A functional claim requires intervention and counterfactual testing. The qualia objection reveals a limit of functionalism but is not a reason to abandon functional insights. Many capacities— attention, memory, control, planning—can be studied fruitfully in terms of functional organisation even if phenomenal character requires an additional account. Do not try to solve the hard problem of consciousness prematurely in Chapter 77. Effective functional access, reportability, discrimination, and subjective feel stand in relations that must be investigated separately; Chapter 78 will focus on that gap. For intentional content, keep open the hybrid possibility of internal role plus external relation. Mental content may involve language, environment, learning history, social norms, and causal role together. Chapter 79 will make this its principal subject. Explanatory pluralism is not ontological carelessness. The same event can admit compatible explanations at the level of reasons, functional organisation, and neural mechanism if their relations are made clear. “Two explanations” does not mean “two substances.” In clinical contexts, behaviour-based screening, neurobiological evidence, and functional impairment are distinct dimensions. Collapsing diagnosis into a single behavioural score or biomarker is epistemically weak and ethically dangerous. In disability contexts, normative behavioural baselines require caution. Treating eye contact, speech cadence, movement, or affect display as universal mental criteria can misclassify neurodivergent or disabled persons. In dialogue with Indian traditions, category translation should proceed in two steps: first understand a term within the source tradition’s own ontology and epistemology; then specify the comparative axis—public sign, internal state, causal relation, continuity, self, or cognition. Ease of translation is not evidence of identity. Mithila’s Navya-Nyāya method is not the historical source of modern functionalism, but its precision in analysing relations, qualifier and qualificand, locus, inferential conditions, and defeaters can be a useful methodological resource for contemporary comparison. The final uttarapakṣa is this: treat behaviour, brain, and function not as three rival objects but as three analytical windows on mental inquiry. When the evidence conflicts, do not conceal the conflict; re-examine the category, the level, and the inference. Indian Dialogue Within the vast traditions of Indian philosophy there is no single modern category corresponding to “mental state.” Nyāya, Vaiśeṣika, Sāṃkhya, Yoga, Buddhist Abhidharma, Mīmāṃsā, Vedānta, and other schools assign different ontologies to cognition, self, mind, body, disposition, and action. Comparison must therefore proceed axis by axis. Nyāya treats cognition as a quality of the ātman, manas as a subtle internal sense, and the external senses and body as standing in distinct relations. Translating this model into behaviourism or identity theory would produce substantial loss. Its questions about inference, signs, and the causal conditions of cognition nevertheless possess comparative value. Nyāya developed ways of recognising the inferential role of behaviour, language, action, and bodily signs in knowledge of other minds. This shows the importance of public evidence without saying that mental state = behaviour. It is therefore inappropriate to call Nyāya an anticipation of logical behaviourism. Navya-Nyāya’s language of relations can offer useful comparative discipline in mental attribution: which state, whose locus, what object or viṣaya, what causal condition, what qualifier, what absence or defeater? Here a dialogue with functional specification is possible, but not an identity. Buddhist Abhidharma analyses mental events or dharmas in causal streams, conditioning relations, and momentary processes. It seeks an account of cognition and action without a stable Cartesian subject. Dialogue with modern functionalism is possible around process-orientation, while doctrine, soteriology, and ontology remain very different. Yogācāra debates over representation and consciousness display the complexity of internal structure. Calling Yogācāra “computational functionalism” would be anachronistic. Comparison becomes meaningful only when specific questions are selected, such as causal dependence, representation, or reflexivity. Sāṃkhya draws a strong distinction between puruṣa and prakṛti, including buddhi and manas, but this is not a literal duplicate of Cartesian mind–matter interaction. Functional mental roles may largely belong to the prakṛtic apparatus while consciousness is connected with puruṣa. This distinctive architecture must be respected. Advaita’s distinction between the empirical mind or antaḥkaraṇa and sākṣin or ātman belongs to a metaphysical project different from modern brain-state identity. The shared phrase “consciousness is fundamental” should not be used to construct a direct equivalence with panpsychism or property dualism. Mīmāṃsā attends to linguistic cognition, action, duty, and testimony. The normative dimension of belief–action relations can therefore be richer than a functional causal chain. This opens a useful dialogue concerning the public relation between mental content and linguistic practice. The available sources for Cārvāka or Lokāyata are often polemical and indirect. Presenting a materialist slogan as an exact anticipation of modern identity theory violates historical caution. Where the sources are limited, writing “unknown” or “disputed” is a rule of the parallel view of history. One methodological strength of classical Indian debate is the pūrvapakṣa–uttarapakṣa structure. From it one may learn to reconstruct behaviourism, identity theory, and functionalism in their strongest forms rather than as ideological caricatures. The pramāṇa framework is not an exact theory of the contemporary scientific method, but it can encourage discrimination among types of evidence. Behaviour, first-person report, neural measures, intervention, testimony, and model output do not carry identical evidential weight; formal discussion of such differences is fruitful. गजेन्द्र ठाकु रक समानान्तर दर्शन — खण्ड २ In modern Indian philosophy, B. K. Matilal, J. N. Mohanty, Jonardon Ganeri, Arindam Chakrabarti, and others have developed critical dialogues between classical categories and contemporary philosophy. The spirit to be followed here is not equivalence but question-sharing. Glorifying claims such as “ancient India had already discovered neuroscience or AI” should be rejected in Indian psychological and philosophical reinterpretation. Historical texts had their own aims, vocabularies, and argumentative environments; modern laboratory evidence operates under a different standard. Parallel maxim: “Indian tradition is not a certificate for modern theory; modern theory is not the measure of Indian tradition. Share questions, keep categories distinct, and test evidence with equal rigour.” Mithila’s Parallel Perspective In Mithila’s philosophical history, the traditions of Nyāya–Navya-Nyāya, Mīmāṃsā, Vedānta, linguistic analysis, and śāstrārtha can enter into dialogue with contemporary questions of mental attribution. But a claim such as “functionalism existed earlier in Mithila” would violate the method of this project. Works such as Gaṅgeśa’s Tattvacintāmaṇi conduct extraordinarily fine-grained analyses of relations concerning knowledge, inference, absence, testimony, and related topics. For theories of mental states, the methodological lesson of this precision— definition, upādhi, vyāpti, defect, counterexample—is important. For example, the statement “this student is not paying attention” moves from a behavioural observation to a mental attribution. A Navya-style discipline would ask: What is the indicator? What is the locus? What alternative causes exist? Are there counter- instances? Is the linguistic criterion clear? Such questions can make educational psychology more careful. Mithila’s multilingual environment—Maithili, Hindi, Nepali, Urdu, Sanskrit, English, and local varieties—shows the limits of behavioural language tests. The inference “less fluent output = less cognition” may be false because of a language barrier. When standardised neuropsychological or educational tools are transferred into the shared India–Nepal cultural region, linguistic norms, literacy, schooling, migration, and socioeconomic context must be examined. Behaviour-based mental categories should be culturally calibrated. In everyday social life, labels such as “quiet,” “mad,” “lazy,” “intelligent,” “shy,” or “eccentric” can turn observable behaviour into a supposed mental essence. Philosophy of mind supplies tools for developing evidential and ethical criteria for such label- based inferences. Caste and occupation create behavioural expectations—whose speech should be deferential, whose mobility is restricted, whose manual skill is described as “natural.” Treating such socially conditioned behaviour as an innate mental type is a social form of behaviouristic essentialism. Gendered behavioural norms can likewise distort mental attribution. An assertive woman may be labelled “aggressive,” a silent woman “modest,” an emotional man “weak.” The same behaviour receives different social readings; public criteria are not automatically neutral. Migration increasingly makes family communication dependent on phones, video, and text. Inferring affection, loyalty, or mental states from response delay, online presence, emoji use, or remittance behaviour requires caution because digital behaviour is evidence stripped of much of its context. During floods or other disasters, stress behaviour changes under survival constraints. Withdrawal, irritability, or impulsive responses should not automatically be read as fixed personality traits or pathology. Situations can temporarily reshape a functional profile. Traditional healing, religious treatment, and biomedical psychiatry offer different causal vocabularies for mental distress. A parallel perspective does not endorse any of them wholesale; it applies standards of safety, outcome, evidence, consent, referral, and dignity. In Mithila’s schools, a practical lesson from behaviourism can be used in reverse for bilingual and multilingual cognitive assessment: measure observable performance, but separately examine confounds such as language proficiency, opportunity, anxiety, sensory disability, and teaching quality. For Digital Maithili AI, functional evaluation cannot consist merely of translating English benchmarks. Code-switching, Tirhuta and Devanagari scripts, morphology, idiom, honorifics, local knowledge, handling of uncertainty, and retrieval provenance are among the functional dimensions that must be tested. Seeing a chatbot’s Maithili fluency and saying that it “understands Maithili” is a behavioural attribution. A stronger functional claim requires tests of generalisation to unseen contexts, correction, memory boundaries, causal intervention, grounding, and sensitivity to error. In brain and mental-health research using population data from Mithila, it would be highly dangerous to conceal the confounding effects of caste, poverty, nutrition, flood exposure, migration, pollution, education, and language and then invent a “regional brain trait.” Identity theory grants no licence for biological essentialism. A Navya-Nyāya-inspired digital annotation scheme could record claim, evidence, locus, relation, qualifier, exception, and uncertainty. In mental-state AI datasets, such structure might reduce overconfident labelling. This is a methodological application, not a claim of historical derivation. Mithila’s parallel maxim: “Observe behaviour, but do not read the mind without observing circumstance; measure the body and brain, but do not reduce the person to a biomarker; test function, but do not mistake fluency for consciousness.” Contemporary Applications In clinical triage, a behaviour-based questionnaire can be useful, but a score is not identical with a disorder. Self-report, duration, impairment, possible medical causes, medication, sleep, trauma, cultural idioms, and clinical judgement must be considered together. In autism assessment, narrow normative criteria concerning eye contact, gesture, or speech style can misclassify people. Functional needs, sensory profile, communication preferences, and developmental history provide the context for behaviour. AI for depression detection may extract risk signals from social-media behaviour, but silence, posting frequency, or word patterns are not direct mental states. False positives, consent, privacy, demographic bias, and crisis-escalation protocols must be addressed. Workplace productivity software that treats keyboard or mouse activity as “focus” or “engagement” is a technological form of crude behaviourism. Knowledge work, offline thinking, disability, caregiving interruptions, and task complexity can undermine the validity of such inferences. Education analytics often uses attendance, clicks, or time-on-page as proxies for learning states. A functional assessment requires richer indicators such as retrieval, transfer, explanation, error correction, and delayed retention. Neuroimaging headlines about a “love centre,” “moral centre,” or “political brain” oversimplify identity theory. The limits imposed by distributed networks, task contrasts, population averages, reverse inference, and replication should be made explicit. In the search for precision-psychiatry biomarkers, correlation, prediction, and causal mechanism must be kept distinct. A biomarker may be useful without establishing the exact neural identity of a mental category. A treatment response to brain stimulation provides evidence from causal intervention, but the conclusion “therefore depression is only a brain disease” would be an overreach. Social and psychological causal pathways can coexist. गजेन्द्र ठाकु र In pain medicine, facial expression, heart rate, behaviour, and physiology are important, but the absence of a patient report is not the absence of pain. Special caution is required with ICU patients, infants, people with dementia or paralysis, and those facing language barriers. Lie-detection technologies may equate behaviour or physiological arousal with deception, but fear, stigma, language, trauma, and interrogation pressure are major confounds. Mental-state inference should not be the sole basis of high-stakes decisions. In AI benchmarking, a chatbot’s human-like answer is a behavioural success. Functional equivalence requires further tests of counterfactual consistency, causal intervention, hidden-state dynamics, long-horizon learning, and grounded action. The Turing Test historically emphasises behavioural evidence for intelligence, but one-shot imitation is not enough for contemporary AI assessment. The testing environment itself can influence the psychology of attribution. Internal activations in a large language model are neural-network states, but the claim “activation pattern = belief” is an identity claim requiring caution. One would need to establish persistence, truth-evaluation, guidance of action, self/world distinction, and counterfactual role. Functionalism can serve as a possible bridge in debates about AI moral status because substrate-neutrality may reduce biological chauvinism. But moral status requires evidence concerning consciousness or valence; treating mere task function as sufficient would be premature. In robotics, the same planning function can be realised in different hardware, a close engineering example of multiple realisability. But the equivalence of engineered function and human mental phenomenology is a separate claim. Brain–computer interfaces can dissociate decoded intention from ordinary motor output. Cases involving locked-in patients sharply expose the limits of pure overt behaviourism: internal neural signals may provide an alternative route for agency and communication. Legal assessments of competence rely on behaviour, interview, records, and cognitive testing; a brain scan alone cannot establish mens rea or capacity. Mental attribution requires evidential pluralism. If an insurance algorithm constructs a mental-risk score from diagnosis codes, medication, or online behaviour, safeguards are required concerning user consent, purpose limitation, explainability, appeal, and discrimination. Consumer “emotion AI” infers sadness, anger, or honesty from facial or vocal patterns. Cross-cultural variability and weak construct validity mean that high-stakes uses should be restricted or subjected to rigorous validation. In generative-AI tutoring, adaptation to a learner’s behaviour demonstrates functional responsiveness. The stronger attribution that the system “understands the student” requires evidence of a learner model, sensitivity to error sources, explanation, uncertainty handling, and longitudinal adaptation. A healthcare robot that detects pain behaviour may perform a useful response function. Whether the robot itself experiences pain is a different question. Blocking this inference from function to experience prepares the ground for Chapter 78. A mental-health chatbot that produces compassionate language may perform part of a therapeutic conversational function. Claims about professional competence, responsibility, crisis detection, privacy, and genuine understanding are separate. Neurodiversity policy can treat functional impairment as relative to environment and provide workplace or school accommodations. Improving person–environment fit instead of searching for one “normal brain type” is a social application of a practical functionalist insight. In human–computer interaction, affordances and user behaviour should be observed, but cognitive load should not be inferred from eye-tracking or clicks alone. Subjective report and task outcome should be triangulated with those measures. In AI safety, phrases such as “the system wants…” may be convenient anthropomorphic shorthand. Operational policy should instead use more precisely testable terms such as goal representation, optimisation pressure, policy behaviour, and learned preference. Contemporary maxim: “The higher the stakes of a mental-state label, the more independently behaviour, mechanism, functional role, context, and uncertainty must be examined.” Chapter Conclusion Behaviourism, identity theory, and functionalism are three decisive turns in modern philosophy of mind. All three react against the Cartesian private substance, but they answer different questions about the criteria, realisation, and causal role of mental states. Behaviourism clarified the public, dispositional, and action-linked aspects of mental vocabulary. Its lasting contribution is not a cult of observability but recognition of public criteria for mental attribution and the importance of context-sensitive behaviour. A pure logical-behaviourist reduction is inadequate because of simulation, suppression, paralysis, contextual dependence, and mental holism. Behaviour is evidence; it is not the exhaustive identity of a mental state. Identity theory preserves the reality of internal mental episodes while advancing the scientific thesis of physical realisation. The Place–Smart–Feigl tradition moved from correlation toward stronger identity claims, but exact type identities require empirical support. Multiple realisability is an effective challenge to strict type identity, not automatically to physicalism. Token physical realisation and higher-level mental kinds may coexist. Functionalism individuates mental kinds through a network of causal roles, internal relations, inputs, and outputs. This provides a powerful framework for psychological generality, computational modelling, and substrate flexibility. The limits of functionalism appear in questions of phenomenal consciousness, content and normativity, embodiment, and excessively abstract organisation. The conclusion “same function = same experience” has not been established; Chapter 78 is therefore necessary. A useful three-level framework is behaviour or public criterion, physical realiser, and functional organisation. In mental inquiry these are often complementary layers of evidence and description rather than enemies. Indian traditions are not “ancestors” of the three modern schools, but they offer independent conceptual resources concerning cognition, sign, self, causal chains, the mind–sense–body relation, and inference. Question-based comparison is the proper form of the parallel method. Mithila’s Navya-Nyāya legacy is not a claim of historical priority but a methodological resource: precision concerning relation, definition, condition, defeater, and inference can be useful in mental-state datasets, education, and AI evaluation. The AI age makes behaviourism newly attractive because machines can speak in human-like ways. But output imitation, functional organisation, semantic content, and conscious experience are distinct questions; success with one does not automatically prove the others. In clinical, educational, legal, and workplace technologies, crude behaviourism or crude neural identity can misclassify people. High-stakes mental inference should be multi-evidential, context-aware, and open to appeal. The chapter’s final maxim is: “Behaviour is a sign of mind, the brain its principal biological realiser, and function the form of its organisation; do not turn sign, realiser, and role into names for one another.”