Relational Frame Theory (RFT) – Steven C. Hayes, Dermot Barnes-Holmes, & Bryan Roche

Human language and higher cognition have long represented the ultimate frontier and, concurrently, the deepest theoretical quandary for empirical behavioral science. Throughout the twentieth century, psychology found itself polarized between two seemingly irreconcilable epistemological camps: a mechanistic behavioral paradigm that struggled to explain the boundless generativity and symbolic abstraction of human speech, and an internalist cognitive revolution that inferred unobservable mental computational architectures at the expense of functional environmental manipulability. The publication of the landmark 2001 treatise, Relational Frame Theory: A Post-Skinnerian Account of Human Language and Cognition, authored by Steven C. Hayes, Dermot Barnes-Holmes, and Bryan Roche, represented an epistemological paradigm shift designed to transcend this historical impasse.

Relational Frame Theory (RFT) proposes that the defining characteristic of human language is not the passive acquisition of formal linguistic representations, nor the statistical extraction of word-frequency associations, nor the unfolding of a genetically pre-programmed Language Acquisition Device. Instead, RFT demonstrates through decades of experimental analysis that human verbal behavior is grounded in a specific, generalized, learned operant performance termed Arbitrarily Applicable Relational Responding (AARR). By learning to relate stimuli under arbitrary contextual control—independent of their immediate formal physical dimensions—human organisms acquire an astonishing capacity to alter the psychological functions of the world without direct physical conditioning.

From its ontological foundation in functional contextualism to its radical reconceptualization of the self, metaphorical reasoning, psychopathology, and fluid intelligence, RFT represents one of the most rigorously tested, empirically grounded, and theoretically unified accounts of complex human functioning in contemporary psychology. This comprehensive analysis will explore the historical genesis, operational mechanics, empirical assessment paradigms, cognitive-developmental implications, and future computational horizons of RFT, establishing its central place in modern behavioral science.

1. Historical and Philosophical Foundations of Relational Frame Theory

1.1 The Evolution from Radical Behaviorism to Functional Contextualism

To fully grasp the magnitude of Relational Frame Theory, one must trace its conceptual ancestry back to B.F. Skinner’s formulation of radical behaviorism. Skinner fundamentally shifted the locus of psychological inquiry from hypothetical internal constructs to functional relationships between an organism’s behavior and its historical and immediate environmental context. Through the three-term contingency—discriminative stimulus ($S^D$), response ($R$), and reinforcing or punishing consequence ($S^R/S^P$)—operant conditioning successfully mapped the environmental selection of overt motor actions across animal species. However, when applied to human verbal behavior, direct contingency accounts repeatedly confronted explanatory boundaries. Mechanistic behaviorism struggled to account for the speed, emergent novelty, and systemic complexity with which children derive symbolic meanings in the absence of explicit, discrete reinforcement histories for every verbal permutation.

Recognizing the mechanistic limits of classical associationism and the reductive tendencies of early behavior-analytic formulations, Hayes and colleagues grounded their work in the philosophical framework of functional contextualism. Drawing directly upon Stephen C. Pepper’s seminal 1942 work on philosophical world hypotheses, RFT departed from formism and mechanism to embrace contextualism as its philosophical home. In contextualism, the root metaphor is the historic act-in-context: an indivisible event that can only be understood through its dynamic situatedness within a specific historical and situational matrix. Unlike descriptive contextualism, which seeks merely to appreciate the rich, idiosyncratic tapestry of ongoing events (as seen in narrative psychology or hermeneutics), functional contextualism is anchored by a precise, pragmatic truth criterion: successful working toward specific, experimentally verifiable analytical goals.

The pragmatic analytical goals of functional contextualism are explicitly defined as the prediction and influence of behavioral events with precision, scope, and depth. “Prediction and influence” are treated not as two independent epistemological operations, but as a single, indivisible criterion; an explanation is functionally true if and only if it points to manipulable variables in the context that allow an experimenter or clinician to both predict and directly alter the behavior under study. The 2001 volume by Hayes, Barnes-Holmes, and Roche codified this philosophical stance, presenting a post-Skinnerian paradigm shift that preserved the radical behaviorist commitment to external environmental accountability while expanding its reach to accommodate the vast generativity of the human symbolic mind.

1.2 Critique and Re-conceptualization of Skinnerian Verbal Behavior

In 1957, B.F. Skinner published Verbal Behavior, an ambitious attempt to extend the principles of operant conditioning to human language. Skinner classified verbal operants according to their functional control, introducing categories such as the mand (verbal behavior under the control of establishing operations and dynamic states of deprivation), the tact (verbal behavior evoked by non-verbal discriminative stimuli through a history of generalized conditioned reinforcement), the intraverbal (verbal responses under the control of antecedent verbal stimuli without point-to-point correspondence), and the autoclitic (behavior that modifies the effects of other verbal operants on the listener). While clinically transformative—particularly in establishing functional communication training for individuals with severe developmental delays—Skinner’s taxonomy treated verbal behavior largely as a collection of vocal or gestural topographies maintained by immediate environmental mediators.

The vulnerability of Skinner’s model was famously seized upon by linguist Noam Chomsky in his devastating 1959 review. Chomsky argued that an operant account reliant entirely on direct conditioning histories could never explain the infinite generativity of syntax or the phenomenon of stimulus poverty—the empirical reality that human infants acquire complex linguistic competence rapidly despite being exposed to fragmented, incomplete, and highly limited samples of linguistic input. While Chomsky leveraged this critique to justify mentalistic nativism, postulating an innate, genetically hardwired Universal Grammar and Language Acquisition Device (LAD), the critique exposed an empirical gap in Skinner’s formulation: human beings systematically derive vast networks of novel verbal relations that have never been directly trained, reinforced, or modeled.

RFT resolved this historical crisis not by retreating to unobservable mental modules, but by fundamentally re-anchoring verbal behavior as generalized learned patterns of relational responding rather than distinct verbal operants tied to specific stimulus-reinforcer histories. Under an RFT reconceptualization, a word does not function merely as an acoustic discriminative stimulus ($S^D$) or a conditioned response ($CR$); it functions as an element within an arbitrarily applicable relational network. By demonstrating that relational responding can be brought under arbitrary contextual control, RFT provided the functional, post-Skinnerian answer to Chomsky: human language is infinitely generative not because of innate mental representations, but because human organisms acquire a generalized, overarching operant class that allows them to relate anything to anything, deriving novel behavioral functions without direct conditioning.

1.3 Ontological and Epistemological Assumptions in RFT

The ontological foundation of Relational Frame Theory is monistic and non-representational. RFT explicitly rejects the dualistic architecture of Cartesian philosophy and cognitive representationalism, which posits an external physical reality mirrored imperfectly by internal mental structures, schemas, or neurological codes. In the RFT paradigm, there is no ontological distinction between the mind and behavior, nor between an internal symbolic representation and an external motor action. Thought, emotion, perception, and overt speech are all conceptualized as activities of the whole organism interacting in and with an historical and current context. They are functional interactions, not static structural entities housed within an intra-psychic container.

Central to this ontology is the continuous, indivisible act-in-context. Psychological events cannot be meaningfully dissected into independent sensory inputs, central cognitive processors, and behavioral outputs without destroying the functional integrity of the event. Even the most private phenomena—silent self-talk, covert visualization, and visceral somatic sensations—are treated as behaviors governed by the exact same physical and functional laws that regulate overt walking, typing, or vocalizing. This monistic view of private events preserves radical behaviorism’s rejection of mentalistic circularity (e.g., asserting that a person avoids social situations “because they have social anxiety,” which mistakes a descriptive private event for an explanatory cause), insisting instead that the causes of behavior must reside in the manipulable, historical context of the organism-environment system.

Epistemologically, RFT evaluates its theoretical adequacy against three strict criteria:

• Precision: The theoretical account must employ a limited set of clearly defined, unambiguous terms that apply consistently across all domains without shifting their referents.
• Scope: The concepts must apply broadly, spanning the entire breadth of human experience from the acquisition of single words in infancy to complex metaphor, cross-cultural ideological conflicts, mathematical deduction, and psychiatric suffering.
• Depth: The formulations must harmonize seamlessly across adjacent scientific disciplines—cohering with biological evolution, developmental psychology, neurobiology, and systemic anthropology without violating physicalist principles.

Through these ontological commitments, RFT constructs a naturalistic, empirical science of human language that avoids mentalism while confronting symbolic cognition directly.

2. The Mechanics of Arbitrarily Applicable Relational Responding (AARR)

2.1 Non-Arbitrary Versus Arbitrarily Applicable Relational Responding

All complex organisms possess the biological capacity for non-arbitrary relational responding. A non-human primate, a rodent, or a pigeon can readily learn to respond to the relationship between physical stimuli based purely on their formal, non-arbitrary dimensions. For example, when presented with a choice between two geometric shapes of varying volumes, an animal can be trained to reliably select the “larger” object, even when entirely novel stimulus pairs are introduced. In this non-arbitrary paradigm, the organism’s behavior is strictly regulated by the physical properties of the stimuli themselves—their comparative size, luminance, acoustic frequency, or chemical concentration. If an experimenter presents a 5-gram metal ball and a 50-gram metal ball, the relation “heavier than” is physically bound to the material reality of those objects.

The definitive evolutionary breakthrough of the human verbal organism is the emergence of Arbitrarily Applicable Relational Responding (AARR). In AARR, the relational responding is no longer bound to, nor governed by, the formal physical attributes of the related events. Instead, the relational performance is brought under the control of socially established, arbitrary contextual cues. Consider a classic economic illustration: a tiny, light-weight piece of paper (a $100 bill) is evaluated as having vastly “greater” value than a heavy, physical nickel coin. Physically, the coin is denser, harder, and heavier; symbolically and relationally, the paper “is more than” the metal. The relation of magnitude has been abstracted away from the physical substrate and placed under the control of arbitrary social conventions.

This capacity for AARR does not materialize out of thin air; it is forged across human ontogeny through vast histories of multiple-exemplar training. Children are systematically taught to respond to stimuli based on cues like “is,” “same as,” “different from,” “bigger,” and “mine,” across thousands of unique contexts. Through this relentless developmental conditioning, the relational activity itself becomes an abstracted, generalized operant class. The child learns to frame events relationally regardless of the physical stimuli involved, ultimately enabling the human nervous system to construct symbolic realities that transcend immediate perceptual constraints.

2.2 Contextual Cues and Their Control Functions

Arbitrarily Applicable Relational Responding operates through the functional interplay of two distinct yet interdependent contextual cues: the relational context, designated as $C_{rel}$, and the functional context, designated as $C_{func}$. These dual cues function as environmental switches that dictate both how stimuli are relationally connected and what psychological functions are subsequently transformed across those relations.

The relational contextual cue ($C_{rel}$) specifies the precise nature or type of relation that must be brought to bear on the targeted events. Examples of $C_{rel}$ include words, symbols, grammatical structures, or spatial arrangements such as “is the same as,” “is opposite to,” “is better than,” “comes before,” or “is inside of.” When an English speaker hears the phrase “Object A is larger than Object B,” the words “larger than” function as the $C_{rel}$, establishing an asymmetrical comparative relational frame. This cue directs the listener to derive a specific relational trajectory: A is greater than B, and automatically, B is less than A, entirely irrespective of whether Object A and Object B actually possess physical size.

Conversely, the functional contextual cue ($C_{func}$) determines which specific psychological functions (e.g., emotional, appetitive, aversive, sensory, motor, or discriminative) are actualized, transferred, or modified across the relational network that has been established by the $C_{rel}$. One might relationally establish that “Brand X is identical to Brand Y” ($C_{rel}$), but this relational network remains psychologically dormant until a $C_{func}$ is introduced. If a trusted source states, “Brand X causes severe allergic reactions,” the $C_{func}$ of biological danger is activated. Because of the established frame of identity, Brand Y instantaneously evokes autonomic distress and behavioral avoidance, even though Brand Y was never physically paired with an allergen or a threat. In daily communication, human beings constantly modulate their behavior by seamlessly navigating and switching between complex arrays of shifting $C_{rel}$ and $C_{func}$ controls.

2.3 AARR as a Generalized Operant Class

In standard behavior analysis, an operant class is defined not by the formal physical topography of the responses, but by its functional relationship to its consequences. For instance, an organism can press a lever with its left paw, its right paw, its chin, or a tool; all distinct topographies fall under the single operant class of “lever-pressing” because they are maintained by the same reinforcing consequence. In RFT, Arbitrarily Applicable Relational Responding is conceptualized as an overarching, generalized operant class. The behavioral act is the *act of relating*, and its functional unity is established and maintained through a rich history of multiple-exemplar training (MET).

During early childhood, the verbal community explicitly trains relational responding across countless distinct physical objects, words, and sensory experiences. A child is shown a real dog and told “dog” (exemplar 1); shown a picture of a dog and told “dog” (exemplar 2); asked to bring the “dog” toy from a basket (exemplar 3); and reinforced when they state that a pug and a wolfhound are both “dogs.” In each exemplar, the specific stimuli differ radically in form, color, sound, and smell, but the functional demand of relating the stimuli in terms of identity remains invariant. Over hundreds of such interactions involving diverse relational forms (sameness, opposition, comparison), the contextual cues ($C_{rel}$) acquire control over the overarching performance of relational framing itself.

Once this generalized relational operant is established, it functions generatively: novel stimuli presented alongside relational cues immediately occasion derived relational responding without requiring explicit training for those specific stimuli. This developmental milestone is virtually unique to humans. While non-human primates can be laboriously trained over thousands of trials to display rudimentary symmetry or transitivities, they fail to demonstrate the fluid, generalized, arbitrarily applicable relational framing that human children exhibit intuitively by the age of three or four.

3. Defining Properties of Relational Frames

3.1 Mutual Entailment

At the architectural core of any relational frame lie three defining functional properties: mutual entailment, combinatorial entailment, and the transformation of stimulus functions. The first of these, mutual entailment, describes the fundamental bidirectionality inherent in relational framing. If an organism is taught or informed that a specific relation exists between a first stimulus (A) and a second stimulus (B) under the influence of a contextual cue ($C_{rel}$), a second relation between B and A is automatically derived or entailed, without any direct training or secondary reinforcement.

Formally, mutual entailment can be represented using the symbolic notation developed in RFT literature:
$$C_{rel} {A \text{ r } B} vdash C_{rel}’ {B \text{ r}’ A}$$
In this formulation, if a contextual cue $C_{rel}$ establishes that stimulus A bears a specific relation ($r$) to stimulus B, this functionally entails ($vdash$) that under an altered or reciprocal contextual cue ($C_{rel}’$), stimulus B bears an entailed relation ($r’$) back to stimulus A.

It is vital to distinguish mutual entailment from Murray Sidman’s foundational concept of symmetry in stimulus equivalence. Symmetry represents a specific, restricted subset of mutual entailment wherein the derived relation is strictly one of equivalence or identity (if A is identical to B, then B is identical to A; $r = r’$). Mutual entailment, however, encompasses asymmetrical, non-equivalent relations. For example, if A is greater than B, the mutually entailed relation is that B is less than A ($r \neq r’$). If A is the father of B, then B is the child of A. Empirical verification protocols rigorously measure both the speed (reaction time latency) and accuracy with which human subjects derive these reciprocal relations, demonstrating that mutual entailment functions as an immediate behavioral response rather than a slow, deliberative syllogistic calculation.

3.2 Combinatorial Entailment

While mutual entailment addresses the bidirectional link between two distinct stimuli, combinatorial entailment describes the systemic integration of three or more stimuli across a relational network. Combinatorial entailment occurs when an organism derives novel relations between two stimuli that have never been directly paired or trained together, based entirely on their mutual relations to one or more intermediary stimuli.

Formally, combinatorial entailment can be stated as follows:
$$C_{rel} {A \text{ r } B} \text{ and } C_{rel} {B \text{ r } C} vdash C_{rel}” {A \text{ r}” C} \text{ and } C_{rel}”’ {C \text{ r}”’ A}$$
If stimulus A bears a relation to stimulus B, and stimulus B bears a relation to stimulus C, the human organism derives relations between A and C, as well as between C and A, governed by the nature of the intervening relational frames.

In standard stimulus equivalence, this is referred to as transitivity (training A=B and B=C yields the derivation of A=C) and equivalence (yielding C=A). In RFT, combinatorial entailment operates across the full taxonomy of relational families. For example:

• Comparative: If X is faster than Y, and Y is faster than Z, it is combinatorially entailed that X is faster than Z, and that Z is slower than X.
• Opposition: If Cold is the opposite of Hot, and Hot is the opposite of Freezing, it is combinatorially entailed that Cold is coordinated with (the same as) Freezing.
• Spatial: If Point Alpha is north of Point Beta, and Point Beta is north of Point Gamma, Alpha is north of Gamma, and Gamma is south of Alpha.

The combinatorial density and flexibility of these networks explain the exponential explosion of human linguistic meaning. By training a tiny fraction of direct relations ($N$), a massive web of derived relations ($N(N-1)$) emerges spontaneously, illuminating the functional basis of cognitive processing speed and linguistic generativity.

3.3 Transformation of Stimulus Functions

Mutual and combinatorial entailment define the structural pathways of relational networks, but the transformation of stimulus functions represents the engine of psychological impact. This property dictates that when a stimulus embedded within an entailed relational network possesses a psychological function—whether respondent, appetitive, aversive, discriminative, perceptual, or reinforcing—that function is dynamically altered, modified, or actualized in other stimuli within the network, in accordance with the specific relational frames and the prevailing functional context ($C_{func}$).

To understand the profound implications of this property, contrast it with classical conditioning. In Pavlovian conditioning, for a neutral stimulus ($CS$) to evoke an autonomic fear response ($CR$), it must be directly and physically paired in temporal contiguity with an unconditioned aversive stimulus ($US$), such as an electric shock. Under RFT, direct pairing is entirely unnecessary. If a person is taught the arbitrary abstract coordinate relations $A = B = C$, and the experimenter subsequently administers an electric shock in the presence of A (making A an aversive stimulus that evokes autonomic arousal, galvanic skin conductance, and tachycardia), the psychological functions of B and C will spontaneously transform. Despite never having been paired with the shock, B and C will immediately evoke fear responses.

Crucially, this transformation is not a crude, uniform generalization; it is modulated precisely by the governing relational frame. If A is established as twice as intense as B, and A is paired with a mild shock, B will evoke a diminished fear response. If A is framed as opposite to B, and A signals an impending reward, B will instantly transform to evoke disappointment or behavioral inhibition. This functional transformation under $C_{func}$ control explains why verbal creatures can experience profound existential despair, panic attacks, or moral ecstasy through thoughts, words, and symbols alone, reacting to abstract verbal constructions as if they were immediate physical crises or triumphs.

4. Taxonomy of Relational Frames

4.1 Coordination, Opposition, and Distinction

Relational framing is not an undifferentiated cognitive capacity; it manifests through distinct families or classes of relational responding, each characterized by its own functional rules for mutual and combinatorial entailment, and its own unique patterns of functional transformation. The foundational triad of relational framing consists of coordination, opposition, and distinction.

Frames of Coordination represent the functional equivalent of “sameness,” identity, or equivalence. Here, stimulus A is framed as identical to or sharing functional equivalence with stimulus B ($A = B$). The relational cue is typically linguistic markers such as “is,” “equals,” “means,” or “is identical to.” Coordination is the ontogenetically earliest and most pervasive frame; it is the basis of naming, labeling, and symbolic representation. When an infant learns that the furry quadruped running through the yard “is a dog,” a frame of coordination is forged between the physical organism, the vocal utterance, and the written word. In coordination, the transformation of stimulus functions is typically symmetric: whatever emotional or semantic functions are acquired by one member transfer directly to the other.

Frames of Opposition involve relating stimuli along a specific, shared continuous physical or conceptual dimension, but positioning them at maximal polar contrariety. Common $C_{rel}$ cues include “is opposite to,” “contrary,” or “reverse.” Unlike simple distinction, opposition requires an underlying dimensional continuum (e.g., temperature, morality, valence). If A is opposite to B along the dimension of heat, and B is opposite to C, A and C are derived as coordinated (hot is opposite to cold; cold is opposite to freezing; hot is similar to freezing in terms of absolute extremity). The transformation of functions across opposition frames is inverse: if A is established as dangerous, an opposite stimulus B may be transformed into a functional signal for safety.

Frames of Distinction are characterized by pure non-identity or difference, governed by cues such as “is not,” “differs from,” or “is distinct from.” Unlike opposition, distinction does not require an organized bipolar continuum. If A is distinct from B, and B is distinct from C, there is no necessary combinatorial derivation regarding the relation between A and C; A and C might be identical, or they might be completely different. Distinction allows the human verbal repertoire to partition the world into categorical non-equivalences, serving as an indispensable precursor to complex classification and exclusion logic.

4.2 Comparison and Hierarchical Relations

As verbal repertoires mature, relational framing expands beyond binary identity into sophisticated quantitative and organizational networks, led by comparative and hierarchical frames.

Frames of Comparison enable the scaling of stimuli along specific quantitative or qualitative dimensional axes. Cues such as “bigger than,” “smaller than,” “faster than,” “smarter than,” “better than,” or “cheaper than” occasion comparative framing. The mathematical and functional entailments are strictly asymmetrical. If $A > B$, then mutually $B < A$. If$A > B$ and $B > C$, then combinatorially $A > C$ and $C < A$. The transformation of stimulus functions across comparative frames is intrinsically non-linear and directional. If an individual learns t\hat Job Offer A pays “more than” Job Offer B, and Job Offer B pays$50,000, Offer A instantly acquires a higher appetitive valence without the person ever seeing the specific numerical salary attached to Offer A. Pathologically, comparative framing forms the psychological engine of self-deprecating social comparison, wherein individuals perpetually frame themselves as “less attractive,” “less successful,” or “less worthy” than arbitrary social targets.

Hierarchical Frames represent one of the most intellectually advanced relational classes, establishing relationships of class inclusion, taxonomical categorization, and part-whole structures. Governed by cues such as “is a type of,” “belongs to,” “contains,” “subsumes,” or “is an attribute of,” hierarchical framing allows for the nesting of specific instances inside superordinate conceptual categories. If an apple is a member of the class “fruit,” and “fruit” is a member of the class “food,” then an apple is combinatorially entailed to be a “food.” Stimulus functions transform downward through the hierarchy with exceptional efficiency: if an individual is told that a rare Amazonian berry belongs to a class of “deadly poisonous fruits,” all avoidant and danger functions associated with the superordinate category are immediately transformed and applied to that specific berry, preserving biological survival through abstract verbal deduction.

4.3 Spatial and Temporal Relational Framing

The human conceptualization of the physical cosmos and the subjective experience of time are profoundly mediated by spatial and temporal relational framing. Rather than reacting exclusively to the sensory immediacy of current environments, spatial and temporal frames allow individuals to cognitively map distant geographies and project behavioral consequences across vast temporal horizons.

Spatial Frames involve topological and directional positioning, regulated by contextual cues like “above,” “below,” “in front of,” “behind,” “inside,” “outside,” “near,” and “far.” Spatial framing provides the structural architecture for physical navigation, architectural design, mechanical engineering, and spatial reasoning. In conjunction with comparative frames, spatial framing enables complex perspective-taking (e.g., “From where you are sitting, the book is to the left of the laptop; from where I am sitting, it is to the right”).

Temporal Frames organize events along a directional, irreversible continuum, governed by contextual cues such as “before,” “after,” “now,” “then,” “yesterday,” “tomorrow,” “sooner,” and “later.” While non-verbal animals live in an experiential present punctuated by associative conditioning and circadian rhythms, temporal framing enables humans to mentally construct an extensive autobiographical past and an infinitely branching simulated future. This capacity carries an immense evolutionary trade-off:

• It allows for preventative planning, agriculture, technological innovation, and deferred gratification.
• It simultaneously acts as the direct psychological conduit for chronic human suffering, allowing individuals to mentally resurrect past traumas through “before” frames (ruminative regret) and fabricate terrifying, non-existent future catastrophes through “then” and “after” frames (anticipatory anxiety).

4.4 Deictic Frames and Perspective-Taking

Perhaps the most profound theoretical achievement of Relational Frame Theory is its functional-behavioral account of human perspective-taking, empathy, and the psychological sense of “self.” This is achieved through the operationalization of Deictic Frames. Unlike coordination, comparison, or spatial frames, deictic frames cannot be anchored to fixed physical properties in the external world. They are fundamentally relational frames that derive their meaning entirely from the vantage point or perspective of the speaker.

The three primary deictic triads identified by RFT are:

• Interpersonal: I versus YOU
• Spatial: HERE versus THERE
• Temporal: NOW versus THEN

Consider the contextual cue “HERE.” There is no physical location on Earth that is inherently “HERE.” “HERE” is simply wherever the speaker happens to be situated at the precise moment of vocalization. As soon as the speaker moves across the room, the former “HERE” instantaneously transforms into a “THERE,” and the new location becomes “HERE.” The same dynamic holds true for “NOW,” which perpetually recedes into “THEN,” and “I,” which transforms into “YOU” when one listens to an interlocutor.

The child acquires deictic framing through exhaustive social interaction, question-and-answer dialogues, and multiple-exemplar perspective-taking training (e.g., “If I were you, and you were me, where would you be sitting?”). This verbal architecture enables individuals to mentally step outside their immediate perceptual locus and adopt the psychological, spatial, and emotional perspective of another person. Empirically, deficits in deictic framing correlate strongly with profound developmental and neuropsychiatric challenges, including Autism Spectrum Conditions (where individuals struggle to derive what another person sees, knows, or feels from “THERE”) and Schizophrenia (where the deictic boundary between “I” and “YOU” breaks down, manifesting as auditory hallucinations or delusions of control).

5. Rule-Governed Behavior and Relational Networks

5.1 Functional Classification of Rules: Pliance, Tracking, and Augmenting

One of the defining applications of RFT to behavioral psychology is its radical refinement of the concept of rule-governed behavior. Skinner initially distinguished between contingency-shaped behavior (actions shaped directly by immediate physical consequences) and rule-governed behavior (actions guided by verbal descriptions of contingencies). RFT deepened this distinction by providing an exhaustive functional classification of rules based on the relational networks that compose them and the specific contingencies that maintain compliance. RFT delineates three functional types of rule-following: pliance, tracking, and augmenting.

Pliance is rule-governed behavior under the primary functional control of socially mediated consequences for compliance or non-compliance with the rule. The rule itself is termed a ply. In pliance, the individual follows the rule not because the behavior naturally produces an intrinsically reinforcing physical outcome, but because the speaker is perceived to be monitoring their behavior and capable of delivering social approval or punishment. An everyday example is a child putting on their winter jacket purely because a parent ordered, “Put your coat on, or you are grounded.” Pliance is essential for early socialization and danger avoidance, but excessive, pervasive pliance engenders an externally focused, submissive behavioral repertoire characterized by severe deficits in autonomy, self-awareness, and personal value discovery.

Tracking is rule-governed behavior under the functional control of the apparent correspondence between the verbal rule (the track) and how the physical or interpersonal environment actually works. In tracking, reinforcement is delivered directly through the natural contingencies of the physical world rather than arbitrary social mediation. For example, following the rule, “If you put fertilizer on the plants, they will grow larger,” is maintained by the visual and biological reality of plant growth. Tracking allows organisms to navigate dangerous or complex terrains without undergoing catastrophic trial-and-error conditioning.

Augmenting refers to verbal rules that alter the reinforcing or punishing potency of previously established consequences, operating functionally as relational establishing or abolishing operations. Augmenting is divided into two subtypes:

• Formative Augmenting: Establishes an entirely new stimulus as a functional reinforcer or punisher through abstract relational coordination (e.g., “If you collect these plastic tokens, you can exchange them for gold medals at the end of the term”).
• Motivative Augmenting: Temporarily amplifies or diminishes the subjective potency of an already established reinforcer by embedding it within a rich relational network of personal values, identity, or historical continuity (e.g., “Drinking this glass of water will not just quench your thirst; it will align you with a healthy, disciplined lifestyle”).

5.2 Insensitivity to Direct Contingencies

While rule-governed behavior confers extraordinary evolutionary advantages—allowing humans to transmit cultural knowledge across generations without every individual having to touch fire to learn it burns—it introduces an alarming behavioral paradox: insensitivity to direct contingencies. Numerous classic behavior-analytic studies (e.g., human performances on Fixed-Interval and Variable-Ratio reinforcement schedules) have demonstrated that while non-verbal animals rapidly adapt their response rates to match changes in underlying reinforcement schedules, humans frequently persist in suboptimal, inefficient, or downright destructive behavioral patterns because they are following an internal, verbally derived rule.

When an individual operates under a rigid verbal rule (e.g., “I must press this button rapidly to get the reward,” or “People will always reject me if I reveal my true feelings”), the derived relational network actively filters out contradictory environmental feedback. The individual stops attending to the direct, non-verbal feedback of the environment. If the direct contingencies change, the person will often rationalize the anomalous outcome (e.g., “I just didn’t press it fast enough,” or “They were just pretending to be kind”), maintaining adherence to the rule despite mounting empirical failure. This verbally mediated contingency blindness is the foundational mechanism underlying psychological rigidity, obsessive-compulsive routines, behavioral perseveration, and depressive traps, wherein individuals act based on what their mind says “should” happen rather than what actually occurs.

5.3 Coherence and the Relational Network Structure

Why do humans so fiercely defend their verbal rules, even at the cost of personal misery and physical inefficiency? RFT answers this question with the construct of relational coherence. From the earliest stages of language acquisition, the verbal community relentlessly trains and reinforces internal consistency, non-contradiction, and logical alignment. A child who says “The dog is a cat” or “Yesterday I will go outside” is immediately corrected. Conversely, the child who constructs coherent, internally consistent relational networks is rewarded with praise, attention, and practical success.

Through this massive socialization history, coherence itself becomes a potent, generalized conditioned reinforcer, while relational ambiguity, contradiction, and incoherence function as profoundly aversive events. Human beings are fundamentally motivated to maintain the structural integrity of their internal relational networks. When confronted with an environmental reality that contradicts a deeply held belief or identity narrative, the organism experiences acute aversive arousal. To resolve this cognitive discomfort (traditionally termed “cognitive dissonance”), individuals will engage in elaborate relational gymnastics: re-framing facts, dismissing contradictory data, vilifying opponents, and inventing narrative justifications—all to preserve the comforting, reinforced state of internal coherence. In clinical contexts, a major objective is to help clients de-literalize this compulsion for coherence, learning to tolerate the perceived incoherence of experiencing painful emotions while pursuing meaningful life actions.

6. The Behavioral Formulation of Self and Perspective

6.1 The Conceptualized Self (Self-as-Content)

One of the most theoretically profound contributions of Relational Frame Theory is its empirical deconstruction of the human “self.” Traditional cognitive and humanistic models often treat the self as an internal substance, a psychic agency, or an elaborate cognitive schema. RFT, by contrast, conceptualizes the self not as a thing, but as an ongoing, layered hierarchy of relational framing behaviors. The most ubiquitous, socially reinforced, and psychologically hazardous of these levels is the Conceptualized Self, often referred to in clinical applications as Self-as-Content.

The conceptualized self is forged through frames of coordination, comparison, and attribution ($C_{rel}$: “I am…”). From childhood, the individual is prompted to verbally classify their traits, capabilities, and liabilities: “I am smart,” “I am uncoordinated,” “I am anxious,” “I am a successful professional,” or “I am permanently damaged.” Over decades, these discrete verbal assessments coalesce into an intricate, densely interconnected relational network—an autobiographical narrative that the person identifies as their essential identity. Because relational coherence is powerfully reinforcing, the individual feels compelled to defend, validate, and remain consistent with this conceptualized story, even when doing so produces immense suffering.

For example, an individual whose conceptualized self is organized around the verbal frame “I am inherently unlovable and broken” will actively interpret neutral social cues as evidence of covert disgust. If someone expresses genuine affection toward them, this positive event threatens the internal coherence of their conceptualized self; consequently, they may push the person away, self-sabotage the relationship, or mentally reframe the love as manipulative pity, all to preserve the coherence of their painful self-narrative. Cognitive fusion with Self-as-Content transforms dynamic living into a desperate, defensive performance of maintaining a scripted verbal identity.

6.2 The Knowing Self (Self-as-Process)

The second functional layer of the self identified by RFT is the Knowing Self, or Self-as-Process. This refers to the continuous, fluid, moment-to-moment verbal tracking of an organism’s own psychological and physiological events in the present. It is operationalized through the dynamic application of deictic and temporal relational frames: “I am feeling a tight sensation in my chest right now“; “I am noticing that I am having the thought that I will fail.”

Self-as-process is the behavioral foundation of mindfulness, metacognition, and emotional intelligence. Unlike the rigid, static definitions of the conceptualized self, the knowing self is characterized by fluid, observational flexibility. It requires the individual to discriminate between their raw visceral experiences (sensations, urges, affective swells) and the verbal evaluations they append to those experiences. Deficits in the knowing self manifest as profound psychological vulnerabilities, including alexithymia (the inability to identify and verbally describe one’s own emotional states), impulsive behavioral acting-out, and dissociative detachment. Cultivating self-as-process allows individuals to consciously register private events as temporary, shifting behavioral phenomena rather than absolute existential truths.

6.3 The Transcendent Self (Self-as-Context)

The third, most sophisticated, and transcendent manifestation of selfhood in RFT is Self-as-Context, alternatively referred to as the Observer Self or the Transcendent Self. Self-as-context emerges as a necessary structural byproduct of mastering the full array of deictic relational frames (I-YOU, HERE-THERE, NOW-THEN). As a human child learns to navigate the constantly shifting vantage points of social interaction, an invariant psychological reality quietly emerges:

The child discovers that their physical body changes over time (I was small THEN; I am tall NOW). Their thoughts mutate ceaselessly (I believed this YESTERDAY; I believe that TODAY). Their feelings fluctuate violently (I was joyful an hour AGO; I am angry CURRENTLY). Their social roles shift depending on the context (I am a student HERE; I am a son THERE). Yet, across every single instance of relational framing throughout their entire lifespan, there is one continuous, invariant constant: the locus or vantage point from which the observing takes place. That locus is always I-HERE-NOW.

Self-as-context is not an entity or a mystical homunculus; it is the functional vantage point, the continuous deictic container within which all psychological events (thoughts, memories, sensations, emotions) arise, shift, and dissipate. When an individual fully differentiates this transcendent perspective from the psychological contents that flow through it, a profound psychological liberation occurs. If I am the context—the canvas upon which the painting appears, the arena in which the battle takes place—then I cannot be harmed or destroyed by the presence of painful thoughts or terrifying anxiety. Fear is simply an event occurring within the I-HERE-NOW space. This functional formulation represents a brilliant convergence between behavior-analytic science and contemplative, meditative traditions, providing an empirical bridge to the experience of pure, unconditioned consciousness.

7. RFT and Cognitive Science: Debates and Paradigmatic Convergence

7.1 Chomskyan Nativism Versus Relational Generativity

For more than four decades, mainstream cognitive science and linguistics operated under the hegemony of Chomskyan nativism. The observation that human infants master language with astonishing velocity, generating an infinite variety of grammatically coherent sentences despite incomplete sensory input, was deemed undeniable proof of an innate, hardwired biological endowment specific to language. Chomsky asserted that general learning principles could never bridge the chasm between raw environmental input and linguistic mastery, cementing the computational metaphor of the mind and marginalizing behavioral science.

Relational Frame Theory delivers an exhaustive empirical and theoretical refutation of this nativist assumption. RFT demonstrates that the human capacity for linguistic generativity is not mediated by an innate syntactic module, but is the direct, predictable outcome of derived relational responding. By establishing that AARR is an overarching operant shaped through multiple-exemplar training, RFT accounts for the explosion of novel verbal derivations without postulating unobservable genetic grammars. If a child learns a handful of basic relational coordinates, they immediately derive exponential networks of unprompted, novel meanings. Computational modeling based on RFT architectures has successfully replicated this emergent generative behavior, demonstrating that artificial networks exposed to multiple-exemplar relational feedback spontaneously begin to derive symmetrical and transitive relations, mimicking the very phenomena Chomsky claimed were impossible under learning theory.

7.2 Symbol Grounding and Embodied Cognition

A classic conundrum in artificial intelligence, cognitive science, and the philosophy of mind is the symbol-grounding problem, famously framed by Stevan Harnad: How do arbitrary physical symbols (e.g., the squiggles on a page or the binary digits in a computer) acquire intrinsic meaning and psychological significance without an infinite regress of secondary symbolic definitions? Cognitive architectures that rely purely on internal computational manipulation of symbols fail to explain how those symbols connect meaningfully to the physical, sensorimotor world.

RFT resolves the symbol-grounding problem from a naturalistic, functional contextualist framework, completely bypassing the need for internal mental representations:

• Symbols are grounded directly through the transformation of stimulus functions under contextual control ($C_{rel}$ and $C_{func}$).
• A symbol (e.g., the word “fire”) is not grounded because it links to an internal mental concept of fire; it is grounded because it is placed in an arbitrary frame of coordination with the physical event of combustion.
• Through that relational frame, the real-world physiological and behavioral functions of fire (heat, pain, visual brightness, avoidance urges) are actualized directly within the symbol itself.

This formulation harmonizes deeply with contemporary theories of embodied cognition, which argue that higher cognitive operations remain inextricably anchored to sensorimotor systems and environmental interactions. In RFT, words do not point inward to a static computational processor; they point outward and functionally tether the organism’s biological responding to an interconnected, socially mediated reality.

7.3 Analogical and Metaphorical Reasoning

Metaphor and analogy have historically presented intractable challenges for behavioral theories, which frequently reduced them to crude stimulus generalization or basic intraverbal associations. Yet, human culture, poetry, scientific breakthroughs, and therapeutic breakthroughs rely fundamentally on the capacity to grasp profound metaphorical parallels between radically dissimilar domains (e.g., “Time is a river,” or “The atom is a miniature solar system”).

RFT provides a mathematically precise, empirically verifiable definition of metaphorical reasoning: metaphor is the relational framing of relational networks, or more simply, relating relations to relations. Consider the classical proportional analogy $A : B :: C : D$ (A is to B as C is to D). For an individual to understand an analogy, they must not merely derive a relation within network 1 ($A \text{ r } B$) and a relation within network 2 ($C \text{ r } D$); they must construct a higher-order frame of coordination between the two derived relational sets:
$$[A \text{ r}_1 B] \text{ Frame of Coordination } [C \text{ r}_2 D]$$

Empirical studies in RFT laboratories have extensively confirmed this model. When subjects are trained on arbitrary, abstract stimulus networks where relation 1 involves comparative scaling and relation 2 involves identical comparative scaling, subjects readily derive a higher-order equivalence between the two networks, demonstrating metaphorical comprehension without explicit instructions. By manipulating the relational complexity, dimensional density, and functional contexts within these networks, RFT researchers can measure and predict the exact velocity with which individuals comprehend complex scientific metaphors, opening unprecedented pathways for accelerating educational pedagogy and therapeutic reframing.

8. Empirical Methodologies and Assessment Protocols in RFT Research

8.1 The Relational Evaluation Procedure (REP)

To experimentally investigate relational framing processes beyond the constraints of traditional discrete-trial Matching-to-Sample (MTS) paradigms, RFT researchers developed the Relational Evaluation Procedure (REP). While MTS successfully demonstrates basic stimulus equivalence (identity matching), it struggles to efficiently assess complex, non-equivalent, multi-dimensional relational networks, such as those involving spatial, temporal, or hierarchical frames with shifting contextual cues.

In a standard REP protocol, participants are presented with an array of stimuli along with specific contextual cues ($C_{rel}$ and $C_{func}$), and are required to evaluate, report on, or interact with the derived relational structures. For example, a computerized REP screen might present a complex, multi-tiered network of arbitrary symbols along with arbitrary cues signifying “greater than,” “opposite to,” and “before.” The participant is then presented with a relational statement (e.g., “Is Symbol X the opposite of Symbol Z?”) and must emit an evaluative response (e.g., selecting “True,” “False,” or indicating an intensity rating). By systematically varying the training pathways, relational distances, and temporal delays, the REP allows experimenters to precisely measure:

• Derivation latency (the millisecond-level cognitive processing speed of the relational derivation),
• Relational flexibility (the ability to rapidly derive new relations when the $C_{rel}$ cue is dynamically inverted), and
• The specific functional parameters governing complex grammatical and syntactic processing.

8.2 The Implicit Relational Assessment Procedure (IRAP)

One of the most consequential methodological innovations arising from the RFT paradigm is the Implicit Relational Assessment Procedure (IRAP), developed by Dermot Barnes-Holmes and colleagues. For years, social and cognitive psychology relied heavily on the Implicit Association Test (IAT) to measure automatic, non-conscious attitudes and social biases. However, the IAT is fundamentally based on associative conditioning principles; it can only measure the simple, unvarnished associative strength or contiguity between two concepts (e.g., “Elderly” and “Bad”), completely failing to assess the *relational nature* of that connection.

The IRAP operates on an entirely different theoretical architecture, measuring relational responding under time pressure. In an IRAP task, participants are presented with pairs of stimuli along with relational contextual cues (e.g., “Similar” versus “Different”, or “True” versus “False”). The software presents alternating blocks of trials:

• Consistent Blocks: The participant is required to confirm socially or relationally dominant pairings under strict response-time limits (e.g., affirming that “Calm” + “Good” = “True”, or “Spider” + “Dangerous” = “True”).
• Inconsistent Blocks: The participant must respond in direct opposition to these dominant relational networks (e.g., affirming that “Calm” + “Good” = “False”, or “Spider” + “Dangerous” = “False”).

The core assumption of the IRAP—formalized in the Relational Elaboration and Coherence (REC) model—is that when a task demands responding that is coherent with an individual’s pre-existing, overlearned relational networks, response latencies are significantly faster. When the task demands responding that contradicts those established networks, brief behavioral hesitations occur as the cognitive system elaborates and attempts to resolve the incoherence, resulting in measurable reaction-time delays (the “IRAP effect”). The IRAP has been successfully utilized across hundreds of empirical studies to measure implicit racial prejudice, implicit self-esteem, deep-seated suicidal ideation, and clinical phobias, consistently demonstrating higher behavioral predictive validity than crude associative measures.

8.3 Neurobiological and Electrophysiological Correlates

Although functional contextualism rejects the reductive claim that psychological events can be fully explained by brain physiology alone, RFT researchers have actively validated relational framing constructs using sophisticated neurobiological and electrophysiological measures, establishing deep interdisciplinary consilience with cognitive neuroscience.

High-density electroencephalography (EEG) and Event-Related Potential (ERP) studies have provided striking neurophysiological corroboration of RFT mechanics. In particular, researchers have examined the N400 wave (a negative-going ERP deflection occurring approximately 400 milliseconds post-stimulus, historically associated with semantic processing anomalies) and the P300 wave (associated with context updating and categorization). RFT experiments demonstrate that when participants are presented with an entailed stimulus that violates a derived relational network (e.g., presenting a stimulus derived as “opposite” when the network demands “coordination”), a robust N400 deflection is immediately evoked. This confirms that derived relational derivations engage the exact same real-time neurofunctional semantic pathways as direct natural language processing.

Furthermore, functional Magnetic Resonance Imaging (fMRI) studies of participants engaged in Arbitrarily Applicable Relational Responding tasks consistently reveal coordinated activation across the dorsolateral prefrontal cortex (dlPFC), the anterior cingulate cortex (ACC), and the posterior parietal cortex. These regions are classic hubs of executive functioning, working memory, and relational integration. Neuroimaging demonstrates that derived relational responding exhibits marked functional plasticity: as individuals undergo intensive relational framing training, these prefrontal activation patterns become increasingly efficient, demonstrating measurable neuroplastic reorganization that parallels gains in fluid cognitive performance.

9. Clinical Applications: The Bridge to Acceptance and Commitment Therapy (ACT)

9.1 The RFT Foundation of the Hexaflex Model

The most famous and clinically revolutionary application of Relational Frame Theory is Acceptance and Commitment Therapy (ACT), developed by Steven C. Hayes and colleagues. While ACT is frequently presented to clinicians through the intuitive, accessible imagery of the “Hexaflex”—a model comprising six core processes of psychological flexibility (Cognitive Defusion, Acceptance, Present Moment Awareness, Self-as-Context, Values, and Committed Action)—every single one of these processes is directly rooted in the mechanics of RFT.

The Hexaflex processes can be systematically translated into relational framing operations:

• Cognitive Defusion: Modifying the functional context ($C_{func}$) to diminish the automatic transformation of aversive stimulus functions, allowing words to be experienced as raw acoustic or visual behaviors rather than literal ontological truths.
• Acceptance: Undoing rigid comparative and temporal relational frames that drive experiential avoidance, fostering willingness to experience aversive private events in the service of chosen values.
• Present Moment Awareness: Enhancing the continuous verbal tracking of immediate physical and private events through fluid Self-as-Process framing (NOW-HERE-I).
• Self-as-Context: Cultivating the transcendent deictic perspective that differentiates the stable locus of observing (I-HERE-NOW) from changing psychological content (THEN-THERE-YOU).
• Values: Constructing hierarchical, motivative augmenting relational networks that imbue ongoing, challenging behavior with positive, verbally derived appetitive functions.
• Committed Action: Building extensive, progressive behavioral chains governed by tracking rules that produce real-world, contingency-shaped reinforcement.

Through this deep theoretical architecture, ACT operates not as an eclectic grab-bag of therapeutic techniques, but as a direct clinical manifestation of fundamental relational science.

9.2 Psychopathology Conceptualized Through an RFT Lens

From an RFT standpoint, the vast majority of non-organic human psychological suffering is an inevitable, predictable byproduct of the evolutionary architecture of human language itself. Language is a double-edged sword: the very relational mechanics that allow humans to build civilizations, cure illnesses, and write poetry concurrently construct the trap of psychopathology.

Psychopathology, within RFT, is primarily driven by three interrelated functional traps:

1. Cognitive Fusion: The complete dominance of derived verbal functions over direct, non-verbal environmental contingencies. When a person is fused with their thoughts, the word “Failure” ceases to be an ephemeral acoustic vocalization and functions psychologically as the *literal physical reality* of being worthless. The individual responds to their internal relational networks as if they were immediate physical threats, losing contact with the rich, corrective feedback of actual sensory experience.
2. Experiential Avoidance: Because of temporal framing (projecting terrifying futures) and causal framing (“I feel bad, *therefore* I cannot live my life”), individuals devote colossal amounts of energy to avoiding, suppressing, or controlling painful private events (thoughts, memories, somatic anxieties). However, because relational networks are generative and interconnected, the very act of trying *not* to think about an anxiety-provoking stimulus embeds that stimulus within a relational frame of coordination with “danger,” paradoxically amplifying its psychological presence and entrenching the individual in a claustrophobic cycle of avoidance.
3. Loss of Flexible Perspective-Taking: Rigid over-identification with the Conceptualized Self (Self-as-Content) creates an extreme brittleness. When a person’s entire psychological existence is anchored to maintaining a rigid self-story, any challenge to that story provokes profound existential terror, leading to severe behavioral constriction, depressive withdrawal, or narcissistic rage.

9.3 Clinical Behavior Analysis in Practice

When an RFT-informed clinician conducts psychotherapy, the therapeutic dialogue is analyzed not merely for diagnostic themes, but as a live, evolving verbal interaction governed by shifting relational frames. The clinician listens closely to the client’s speech to map the underlying $C_{rel}$ and $C_{func}$ dynamics: Is the client using rigid comparative frames (“I am weaker than everyone else”)? Are they trapped in unhelpful temporal frames of regret or catastrophizing? Are their actions maintained by toxic pliance to social expectations, or by flexible tracking of their own life values?

To dismantle cognitive fusion and soften destructive rules, the therapist strategically alters the functional context ($C_{func}$) rather than arguing with the client’s thoughts. Directly arguing with a client (e.g., trying to logically dispute a catastrophic thought) is often clinically counterproductive because engaging in logical debate simply builds more relational networks around the toxic thought, reinforcing its psychological gravity. Instead, the clinician uses defusion techniques—such as having the client repeat a dreaded word aloud for sixty seconds until it loses its meaning, or singing an anxious thought to the tune of “Happy Birthday.” These techniques do not change the relational structure (the thought still says what it says), but they radically dismantle the $C_{func}$, stripping the thought of its power to evoke autonomic panic.

Furthermore, therapists systematically deploy metaphors and experiential exercises. Because metaphors relate one relational network to another, they bypass the client’s rigid literal rules, allowing the client to view their life struggles from an external perspective and derive novel behavioral solutions. The deliberate training of deictic perspective-taking allows clients to step back into the transcendent sanctuary of Self-as-Context, cultivating profound self-compassion and psychological flexibility across the entire spectrum of clinical disorders—from major depression and obsessive-compulsive disorder to chronic physical pain and complex trauma.

10. Developmental and Pedagogical Applications of RFT

10.1 Relational Training Programs for Neurodevelopmental Disorders

One of the most compassionate and empirically validated triumphs of Relational Frame Theory has been its application to developmental psychology, specifically in designing targeted interventions for children with Autism Spectrum Disorder (ASD) and other neurodevelopmental delays. Historically, early intensive behavioral interventions (EIBI) rooted in traditional applied behavior analysis (ABA) excelled at teaching direct operants—such as discrete mands, basic tacts, and rote receptive labeling. However, clinicians frequently observed that these children hit a developmental ceiling: they struggled profoundly with linguistic generativity, symbolic play, empathy, and the derivation of untrained relations.

RFT diagnosed the precise source of this developmental impasse: traditional discrete-trial training was teaching individual verbal topographies, but it was failing to teach the generalized operant class of Arbitrarily Applicable Relational Responding. In response, RFT researchers developed explicit, step-by-step Relational Training Protocols. Using errorless learning, visual scaffolding, and systematic multiple-exemplar training (MET), children are taught to derive mutual entailment, combinatorial entailment, and the transformation of functions across diverse relational families.

The pedagogical sequence begins with simple symmetrical coordination (matching real objects to pictures, pictures to sounds, sounds to printed words), rapidly moves through comparative frames (teaching “more” versus “less” using arbitrary tokens), and culminates in intensive deictic perspective-taking training. Children are trained across hundreds of exemplars to answer questions that systematically shift the I-YOU, HERE-THERE, and NOW-THEN axes (e.g., “I have a red ball and you have a blue ball. What do I have? If I were you, what would you have?”). Empirical studies have demonstrated that this systematic relational training not only establishes functional language in previously non-verbal or minimally verbal children, but remediates fundamental Theory of Mind (ToM) deficits, enabling authentic social empathy and flexible communication.

10.2 The PEAK Relational Training System

The translation of RFT into a standardized, comprehensive developmental curriculum reached its institutional zenith with the creation of the PEAK (Promoting the Emergence of Advanced Knowledge) Relational Training System, developed by Mark R. Dixon and his colleagues. PEAK represents a monumental evolution in behavioral assessment and curriculum design, bridging the gap between traditional Skinnerian verbal operants and post-Skinnerian relational generativity.

The PEAK curriculum is systematically structured into four interconnected, developmentally progressive modules:

1. Direct Training Module: Focuses on foundational, contingency-shaped behavioral repertoires using traditional discrete-trial conditioning (basic mands, tacts, motor imitation, and primary match-to-sample skills).
2. Generalization Module: Systematically trains the broad generalization of established behavioral topographies across novel stimuli, untrained environments, and unfamiliar instructors.
3. Equivalence Module: Introduces the systematic training and testing of stimulus equivalence classes (reflexivity, symmetry, and transitivity), initiating the emergence of derived symbolic networks.
4. Relational Training Module: The full clinical embodiment of RFT. This module systematically assesses and trains the entire taxonomy of relational frames—coordination, opposition, distinction, comparison, spatial, temporal, hierarchical, and deictic—across complex, arbitrarily applicable stimulus environments.

The psychometric properties of PEAK have been extensively validated across dozens of peer-reviewed empirical studies. Performance on the PEAK assessment correlates remarkably well with standard psychometric intelligence tests, such as the Wechsler Intelligence Scale for Children (WISC) and the Stanford-Binet Intelligence Scales. Large-scale longitudinal implementations in public school districts and specialized educational clinics have demonstrated that children undergoing structured PEAK relational training achieve unprecedented linguistic derivations, substantially closing the cognitive achievement gap and fostering authentic behavioral independence.

10.3 Enhancing Fluid Intelligence and Academic Competence

For more than a century, mainstream psychometrics treated fluid intelligence ($Gf$)—the capacity to reason abstractly, solve novel problems, and identify underlying patterns independent of acquired cultural knowledge—as a largely immutable, genetically determined biological trait. While crystallized intelligence ($Gc$) could be expanded through education, fluid intelligence was presumed to be biologically fixed, displaying high heritability and resisting intentional educational interventions.

Relational Frame Theory has radically challenged this fatalistic dogma. RFT posits that what the psychometric literature measures as “fluid intelligence” is not an immutable biological engine, but is, in large part, the speed, density, and flexibility of an individual’s Arbitrarily Applicable Relational Responding repertoire. Abstract reasoning tests (such as Raven’s Progressive Matrices or standard matrix analogies) are, at their core, tests of derived relational framing—demanding that the test-taker derive complex comparative, spatial, and coordinate relations among arbitrary visual geometric arrays.

To put this theoretical hypothesis to the ultimate empirical test, Bryan Roche and his research team developed the SMART (Strengthening Mental Abilities with Relational Training) program. SMART is an intensive, automated, computerized cognitive training intervention that trains children and adolescents through thousands of rigorous, multi-tiered relational framing tasks, specifically targeting arbitrary comparative and equivalence relations under stringent feedback. The empirical results have sent shockwaves through the educational and psychological communities:

• Multiple controlled studies published by Cassidy, Roche, and colleagues have demonstrated sustained, statistically significant increases in full-scale IQ scores—ranging from 10 to over 25 IQ points—following the completion of the SMART protocol.
• These cognitive gains were not confined to the trained tasks; they generalized robustly to standardized academic achievement metrics, yielding marked improvements in mathematical reasoning, reading comprehension, and spatial problem-solving.

By demonstrating that intelligence can be systematically trained as a behavioral repertoire of relational responding, RFT has unlocked revolutionary horizons for public education and social equity.

11. Contemporary Advancements: The MDML and HDML Frameworks

11.1 The Multi-Dimensional Multi-Level (MDML) Framework

As Relational Frame Theory entered its second decade of intensive empirical investigation, researchers recognized that the original conceptualization of relational framing—treating relational frames largely as discrete, categorical operational types—was insufficient to capture the dynamic, non-linear, and continuous nuances of complex human language. To address this need for greater analytical granularity, Dermot Barnes-Holmes, Yvonne Barnes-Holmes, and their collaborators formulated the Multi-Dimensional Multi-Level (MDML) Framework.

The MDML framework constructs a comprehensive, 16-intersection analytical matrix by crossing four distinct levels of relational responding with four operational dimensions:

Levels of Relational Responding	Dimensions of Analysis
1. Mutual Entailing: The derivation of a single bidirectional relation between two stimuli.	A. Coherence: The degree to which the relational responding aligns with the broader relational network and normative social context.
2. Combinatorial Entailing: The integration of three or more stimuli across an interconnected relational web.	B. Complexity: The density, multi-dimensionality, and organizational sophistication of the stimuli and relations involved.
3. Relating Networks: Deriving relations between entire relational networks (the foundation of metaphorical and analogical reasoning).	C. Derivation: The historical distance of the responding—ranging from initial, effortful derivation to overlearned, automatic derivation.
4. Relating Relations: The highest-order framing, involving abstract hierarchical reflection on relational framing processes themselves.	D. Flexibility: The ease with which the relational performance can be modified, updated, or reversed under shifting contextual demands.

By mapping any behavioral act onto this 16-intersection matrix, researchers and clinicians can assess human cognitive functioning with extraordinary precision. A child might display high coherence and low complexity at the level of mutual entailing, while an adult suffering from severe OCD might demonstrate hyper-coherent, hyper-complex relational networks that suffer from profound deficits in relational flexibility. The MDML framework moved RFT away from static, binary categorizations, providing a continuous, dynamic topology of human cognition.

11.2 The Hyper-Dimensional Multi-Level (HDML) Framework

Building directly upon the MDML matrix, the Hyper-Dimensional Multi-Level (HDML) Framework represents the current cutting edge of relational frame science. Developed to forge a seamless, unbreakable bridge between advanced cognitive linguistics and the radical behaviorist imperative for immediate functional manipulability, the HDML explicitly incorporates the motivational, orienting, and emotional dimensions of psychological events directly into the relational matrix.

The conceptual core of the HDML is the ROE-M (Relational, Orienting, and Evoking Multi-level) behavioral formulation. In this model, every psychological event is conceptualized as an indivisible triad:

• Relational Responding (R): The specific arbitrarily applicable relational framing performance taking place across the MDML levels.
• Orienting Functions (O): The immediate attentional and sensory capture of the organism by the stimuli embedded in the network.
• Evoking Functions (E): The specific appetitive, aversive, or emotional behavioral tendencies that are transformed and actualized through those relations.

By integrating the Orienting and Evoking functions directly into the multi-level relational analysis, the HDML reconciles the historical, forty-year divide between traditional applied behavior analysis (which focused almost exclusively on immediate evocative and reinforcing functions) and modern cognitive-behavioral theory. Current empirical studies are actively demonstrating that the HDML possesses unmatched predictive power in modeling how real-time emotional reactivity interferes with complex decision-making, ideological radicalization, and psychiatric relapse.

11.3 Implications for Artificial Intelligence and Natural Language Processing

The dramatic emergence of Large Language Models (LLMs)—such as GPT-4 and its successors—has ignited ferocious debates across cognitive science, linguistics, and philosophy regarding whether deep learning neural networks possess authentic language comprehension. While LLMs exhibit breathtaking linguistic fluency, generating contextually coherent prose across any domain, RFT provides an incisive, scientifically grounded framework for understanding both the capabilities and the profound limitations of statistical artificial intelligence.

LLMs are fundamentally massive statistical prediction engines. They operate through deep layers of transformer architectures that compute high-dimensional mathematical associations and token probabilities based on vast corpora of human text. From an RFT perspective, LLMs possess extraordinarily dense, complex webs of associative and intraverbal relations; however, they do not engage in authentic Arbitrarily Applicable Relational Responding. Crucially, an LLM lacks:

• An embodied, situated biological history in an ongoing physical context.
• A genuine functional context ($C_{func}$): the words within an LLM have no intrinsic emotional, appetitive, or respondent functions; they are mathematical weights.
• True transformation of stimulus functions: an LLM can generate the sentence “The room is filling with lethal carbon monoxide,” but it cannot experience the autonomic orientation, somatic terror, or behavioral flight that defines the functional transformation of that relational network in a living organism.

RFT theorists and AI researchers are currently collaborating to design novel, hybrid cognitive architectures. By infusing symbolic neural networks with explicit RFT mechanics—programming artificial systems to explicitly derive mutual entailment, combinatorial entailment, and the transformation of functions under shifting contextual cues—computer scientists aim to advance artificial intelligence beyond statistical mimicry toward authentic, grounded symbolic cognition.

12. Epistemological Critiques, Controversies, and Future Horizons

12.1 Critiques from Within Behavior Analysis

Despite its profound successes, Relational Frame Theory has not traversed its academic trajectory without intense scientific controversy, experiencing some of the fiercest resistance from within its own intellectual home: the traditional behavior-analytic community. Conservative Skinnerian purists, often self-identified as classical radical behaviorists or structural verbal behavior analysts, have repeatedly leveled sharp conceptual criticisms against RFT.

The primary critique centers on the charge of hypothetical mentalism and linguistic opacity. Traditionalists argue that terms such as “mutual entailment,” “combinatorial entailment,” and “relational frames” introduce an unnecessary, obfuscating layer of theoretical vocabulary that threatens to detach behavior analysis from its parsimonious roots in the three-term contingency. Some critics have asserted that AARR is not a novel, fundamental behavioral primitive at all, but merely a complex combination of simpler, already established operant processes, such as conditional discriminations, mediated stimulus equivalences, and generalized imitation.

RFT theorists have mounted extensive, empirically rigorous counter-arguments to these objections. Decades of laboratory studies have demonstrated that derived relational responding exhibits functional properties that cannot be accounted for by chaining or conditional discrimination alone. RFT researchers point out that Skinner himself continually called for a functional, experimental analysis of complex behavior; clinging to Skinner’s 1957 taxonomy as an inviolable sacred text violates the core progressive spirit of radical behaviorism. RFT does not invent unobservable mental agents; its concepts describe directly observable, experimentally manipulable interactions between the organism and arbitrary environmental cues, preserving the physicalist, monistic integrity of behavioral science while vastly expanding its analytical reach.

12.2 Methodological and Replicability Challenges

Beyond theoretical disputes, RFT has faced significant methodological and replicability scrutiny, a necessary crucible for any mature scientific paradigm. One prominent challenge is the sheer labor-intensive, technically demanding nature of RFT laboratory protocols. Conducting experimental studies on complex relational derivations requires sophisticated computerized software, extensive pre-training phases, hundreds of baseline trials, and meticulous control over extra-experimental linguistic histories. Consequently, early empirical studies were often characterized by relatively small sample sizes, raising questions regarding the broad generalizability and statistical power of some findings.

A second major methodological battleground has centered on the Implicit Relational Assessment Procedure (IRAP). While the IRAP was hailed as an innovative tool for measuring derived relational responding, critics and independent researchers have highlighted significant challenges regarding its psychometric reliability. The IRAP effect has sometimes exhibited low test-retest reliability, high participant attrition rates (as subjects frequently fail to meet the rigorous speed and accuracy criteria during inconsistent blocks), and complex order effects. Barnes-Holmes and colleagues have actively addressed these critiques through the formulation of the REC and HDML models, refining the scoring algorithms and experimental architectures of the task.

Additionally, the dramatic claims of massive IQ gains achieved through the SMART relational training program have naturally invited demands for large-scale, independent, preregistered replications. While numerous studies have replicated the cognitive benefits of relational training, independent laboratories have called for massive, multi-site randomized controlled trials with active control groups to definitively confirm the effect sizes and establish the long-term educational durability of these cognitive enhancements.

12.3 The Horizon of Relational Frame Theory

As Relational Frame Theory moves deeper into the twenty-first century, its research horizon is expanding far beyond the walls of basic psychological laboratories, reaching into the grandest systemic challenges confronting human civilization. One of the most urgent frontiers is the application of RFT to large-scale cultural phenomena, ideological polarization, and digital socio-political dynamics. In an era dominated by algorithmic echo chambers and intense tribal animosity, RFT provides a precise functional analysis of how symbolic networks become weaponized:

Social media algorithms are exquisitely engineered to reinforce insular, highly coherent relational networks. When opposing political groups become framed through rigid relations of opposition and negative moral comparison, any contradictory empirical fact is automatically rejected to preserve relational coherence. RFT researchers are actively modeling how cultural narratives, systemic prejudice, and institutional propaganda can be functionally deconstructed, designing societal-level interventions that promote perspective-taking and psychological flexibility on a cultural scale.

Furthermore, RFT is forging profound multidisciplinary integrations with evolutionary biology and multi-level selection theories, collaborating with evolutionary thinkers like David Sloan Wilson to demonstrate how symbolic relational systems function as a major evolutionary transition in human inheritance. Simultaneously, the proliferation of digital mental health technologies and mobile application frameworks is enabling the deployment of automated, algorithmically tailored RFT and ACT interventions to millions of individuals globally, democratizing access to evidence-based psychological care.

The legacy forged by Steven C. Hayes, Dermot Barnes-Holmes, and Bryan Roche is nothing short of a fundamental re-imagining of human nature. By daring to look into the magnificent, terrifying complexity of the symbolic mind through the lens of empirical behavioral science, Relational Frame Theory has forever dissolved the artificial wall between behaviorism and cognition. It stands today as a testament to the fact that the boundless generativity of human language, the deepest depths of human suffering, and the highest peaks of human consciousness can all be understood, honored, and healed through the naturalistic science of the contextual act.

Conclusion

Relational Frame Theory represents one of the most intellectually ambitious, methodologically rigorous, and therapeutically transformative undertakings in the history of behavioral science. By identifying Arbitrarily Applicable Relational Responding as the core behavioral engine of human language and higher cognition, RFT successfully bridged the historic divide between radical behaviorism and cognitive science. It rescued behavioral psychology from the mechanistic limitations of simple associative conditioning without succumbing to the unobservable, mentalistic architectures of cognitive representationalism.

Through its mathematically precise operationalization of mutual entailment, combinatorial entailment, and the transformation of stimulus functions, RFT decoded the functional mechanics that make human language infinitely generative. It exposed the psychological roots of rule-governed insensitivity, demystified the behavioral formation of the self across its conceptualized, processual, and transcendent dimensions, and provided an empirical foundation for modern clinical and educational interventions. From the clinical triumphs of Acceptance and Commitment Therapy to the transformative educational curricula of PEAK and SMART, RFT has demonstrated that understanding the relational nature of human thought directly empowers us to alleviate suffering and elevate human intellectual potential.

As psychology, artificial intelligence, and evolutionary science continue to converge, Relational Frame Theory offers a unifying, functional contextualist blueprint for the future. It reminds us that human beings are neither passive biological automata nor disembodied computational machines; we are historical, verbal organisms continually weaving webs of symbolic meaning across context and time. In mastering the science of how we frame our world, we ultimately acquire the transformative capacity to reframe our shared human future.

References

• Barnes-Holmes, D., Barnes-Holmes, Y., Hussey, I., & Luciano, C. (2016). Relational Frame Theory: Finding its historical and intellectual roots and reflecting upon its future development. In R. D. Zettle, S. C. Hayes, D. Barnes-Holmes, & A. Biglan (Eds.), The Wiley Handbook of Contextual Behavioral Science (pp. 115-128). Wiley-Blackwell. https://doi.org/10.1002/9781118489857.ch8
• Barnes-Holmes, D., Finn, M., McEnteggart, C., & Barnes-Holmes, Y. (2018). Derived stimulus relations and the high-dimensional multi-level (HDML) framework: Extending the challenge to cognitive models of complex human behavior. Behavior Modification, 42(2), 155-173. https://doi.org/10.1177/0145445517743705
• Cassidy, S., Roche, B., & Hayes, S. C. (2011). A relational frame training intervention to raise intelligence quotients: A pilot study. The Psychological Record, 61(2), 173-198. https://doi.org/10.1007/BF03395755
• Chomsky, N. (1959). A review of B. F. Skinner’s Verbal Behavior. Language, 35(1), 26-58. https://doi.org/10.2307/411334
• Dixon, M. R., Belisle, J., Whiting, S. W., & Rowsey, K. E. (2014). PEAK Relational Training System: Direct Training Module. Shawnee Scientific Press.
• Harnad, S. (1990). The symbol grounding problem. Physica D: Nonlinear Phenomena, 42(1-3), 335-346. https://doi.org/10.1016/0167-2789(90)90087-6
• Hayes, S. C., Barnes-Holmes, D., & Roche, B. (Eds.). (2001). Relational Frame Theory: A Post-Skinnerian Account of Human Language and Cognition. Kluwer Academic/Plenum Publishers. https://doi.org/10.1007/b108413
• Hayes, S. C., Strosahl, K. D., & Wilson, K. G. (2012). Acceptance and Commitment Therapy: The Process and Practice of Mindful Change (2nd ed.). Guilford Press.
• Pepper, S. C. (1942). World Hypotheses: A Study in Evidence. University of California Press.
• Roche, B., Cassidy, S., & Stewart, I. (2013). Relational Frame Theory: A contemporary behavior-analytic account of language and cognition. In G. J. Madden (Ed.), APA Handbook of Behavior Analysis (Vol. 1, pp. 181-200). American Psychological Association. https://doi.org/10.1037/13937-007
• Sidman, M. (1994). Equivalence Relations and Behavior: A Research Story. Authors Cooperative.
• Skinner, B. F. (1957). Verbal Behavior. Appleton-Century-Crofts. https://doi.org/10.1037/11256-000
• Törneke, N. (2010). Learning RFT: An Introduction to Relational Frame Theory and Its Clinical Application. New Harbinger Publications.