If you’ve spent any time in ABA, you already know how to teach a kid to label a cup. You hold up the cup, say “cup,” reinforce when they echo it, and over enough trials you’ve got a tact. Skinner figured that out in 1957 and it works. But here’s the question that’s bothered behavior analysts for sixty-plus years: how does a four-year-old, after learning that a “cup” is the thing they drink from, suddenly point to a cup in a picture book they’ve never seen, ask for a cup at grandma’s house, and refuse a “cup” when it’s actually a mug? Nobody trained any of that. The relations between word, object, picture, and category emerged on their own. Relational Frame Theory is behavior analysis’s attempt to explain that emergence — and to do it without abandoning the operant conditioning framework that makes ABA work in the first place.
This guide walks through RFT the way we wish someone had walked us through it: friendly, clinically grounded, honest about where it gets debated, and pointed straight at the work you actually do with clients. We’ll cover the core mechanics (mutual entailment, combinatorial entailment, transformation of stimulus functions), the family of relational frames, the deictic frames behind perspective-taking, and how all of this scaffolds modern verbal behavior programs like PEAK and T-FT. We’ll also tie RFT to Acceptance and Commitment Therapy (ACT), because the two were built by the same people for the same reason. By the end you should be able to read an RFT paper without bouncing off the jargon and know whether RFT-based teaching belongs in your next program.
What Is Relational Frame Theory?
Relational Frame Theory (RFT) is a behavior-analytic account of human language and cognition. It was developed primarily by Steven C. Hayes, Dermot Barnes-Holmes, and Bryan Roche, with the canonical statement appearing in their 2001 textbook Relational Frame Theory: A Post-Skinnerian Account of Human Language and Cognition. The theory’s core claim is straightforward: a huge amount of human verbal behavior consists of derived stimulus relations — relations between stimuli that the learner was never directly trained on. RFT calls the act of relating one stimulus to another a relational response, and it argues that these relational responses are themselves a generalized operant. In plain English: humans learn how to relate things, and once they learn that skill, they apply it to brand-new content without further training.
Three features define a relational frame: it is mutually entailed (relating A to B implies a relation from B to A), it is combinatorially entailed (relations between pairs of stimuli combine into new relations across a network), and it produces a transformation of stimulus functions (the psychological function of one stimulus changes based on its relation to others). If a behavior shows those three features and is brought under contextual control by cues like “same as,” “opposite of,” or “bigger than,” it counts as relational framing.
Why Skinner’s Verbal Behavior Wasn’t Enough (the “derived” problem)
Skinner’s Verbal Behavior (1957) gave us mands, tacts, echoics, intraverbals, and a powerful framework for teaching language as operant behavior. For non-verbal learners and early language programs, it’s still the backbone of what we do. But the framework was built around directly trained verbal operants. If a learner only emits responses that were directly reinforced, the model holds. The trouble starts when learners emit responses that nobody trained.
This is the “derived problem,” and the cleanest demonstration is stimulus equivalence, Murray Sidman’s work from the 1970s and 80s. Teach a child that spoken “dog” goes with a picture of a dog (A→B), and that the picture goes with the printed word “dog” (B→C), and most typically developing children will spontaneously match the printed word back to the spoken word (C→A), the picture back to the spoken word (B→A), and so on. Nine relations emerge from training two. Sidman called this an equivalence class. He treated it as a basic stimulus phenomenon — important, but not fully explained.
RFT picks up where Sidman left off and says: equivalence is just one kind of derived relation, and the reason humans do this so easily is that we’ve learned relating itself as a generalized operant. Once a child has a history of being reinforced for putting things into “same as” relations across hundreds of exemplars, they apply that frame to any new pair of stimuli a caregiver presents. The frame is the operant. The content is interchangeable.
The Core Concept: Mutual Entailment
Mutual entailment is the simplest piece of relational responding. If a learner is trained that A relates to B in some way, mutual entailment says the learner will derive the corresponding relation from B back to A without further training.
- If A is the same as B, then B is the same as A.
- If A is bigger than B, then B is smaller than A.
- If A is the opposite of B, then B is the opposite of A.
The key word is derived. The B→A relation was never directly reinforced. The learner produces it because they have a long history of “same,” “bigger,” and “opposite” being used in symmetrical, asymmetrical, or oppositional ways across many exemplars. The contextual cue (“same,” “bigger”) tells them which relation to apply. RFT calls this arbitrarily applicable relational responding: the relation is determined by the social-verbal context, not by the physical properties of the stimuli. A penny and a dime are physically different, but if you say “a penny is worth less than a dime,” the child relates them on a comparison frame even though the penny is bigger.
Combinatorial Entailment and Transformation of Stimulus Functions
Combinatorial entailment is what happens when you put two mutual entailments together. Train A→B and B→C, and the learner derives A→C and C→A. Train A is bigger than B and B is bigger than C, and the learner derives A is bigger than C (and C is smaller than A). This is how relational networks expand exponentially — a handful of trained relations produces a web of derived ones.
The third feature, transformation of stimulus functions, is where RFT gets clinically interesting. The psychological function of a stimulus — what it makes you feel, do, or approach — changes based on its position in a relational network. The textbook example: pair the nonsense word “vug” with a mild shock so it elicits aversion. Then teach the learner that “vug” is the opposite of “zog.” Without ever pairing “zog” with anything, the learner now treats “zog” as appetitive or safe. The function transferred along the relational frame.
Now translate that into a clinic. A child who has been bullied has a long network of “kids at school = scary.” A teacher introduces a new peer and says, “He’s just like the friends you have at the playground.” If “playground friends” is in a safety frame, the new peer’s function transforms. If the network instead links school to a fear frame, the same introduction might transform the playground peer the other direction. This is why RFT folks insist that language isn’t neutral. Words drag function across networks, and the networks are built from a person’s verbal history.
The Family of Relational Frames
RFT identifies a family of relational frames — distinct patterns of relating that humans learn as generalized operants. Each frame has its own contextual cues, its own entailment structure, and its own clinical implications. Here’s the quick map:
| Frame | Contextual cues | What it lets the learner do | Clinical relevance |
|---|---|---|---|
| Coordination (same as) | “is,” “same,” “equals” | Treat stimuli as equivalent | Receptive/expressive labels, tacts, equivalence-based instruction |
| Distinction (different from) | “different,” “not the same” | Treat stimuli as non-equivalent | Discrimination training, category sorting |
| Opposition (opposite of) | “opposite,” “anti-“ | Reverse the function | Antonyms, emotional reframing |
| Comparison (more/less) | “bigger,” “faster,” “more” | Order stimuli along a dimension | Quantitative reasoning, preference scaling |
| Hierarchical (part/whole) | “is a kind of,” “contains” | Build categories and class membership | Self-as-context work in ACT, taxonomies |
| Temporal (before/after) | “before,” “after,” “during” | Order events in time | Schedules, sequencing, future-oriented behavior |
| Causal (if/then) | “because,” “if,” “then” | Relate events as cause and effect | Rule-governed behavior, contingency learning |
| Deictic (perspective) | “I/you,” “here/there,” “now/then” | Take a perspective | Theory of mind, empathy, self-awareness |
You don’t have to memorize this. You do need to recognize that when you’re teaching “the dog is bigger than the cat but smaller than the horse,” you’re recruiting a comparison frame, not a coordination frame, and your instructional logic should match. Programs like PEAK and T-FT explicitly target specific frames in specific sequences for exactly this reason.
Deictic Frames and Perspective-Taking
The deictic frames deserve their own section because they are how RFT explains perspective-taking — what mainstream psychology calls theory of mind. There are three deictic relations:
- I / You (interpersonal perspective)
- Here / There (spatial perspective)
- Now / Then (temporal perspective)
These relations are unique because they can only be learned through the speaker’s own perspective. You can’t point to “I” the way you can point to “cup.” The learner has to abstract the relation from countless exemplars: when mom says “I want the cup,” I is the speaker; when I say “I want the cup,” I is me. Over time the learner derives a stable self-perspective and the ability to flexibly shift into another person’s perspective (“If I were standing where you are standing, the cup would be on my right”).
Deficits in deictic framing predict deficits in perspective-taking tasks. Research by Louise McHugh and colleagues, building on the Barnes-Holmes lab, has shown that explicit training in I/you, here/there, and now/then relations can improve perspective-taking performance in autistic learners and in learners with intellectual disabilities. This is one of the cleaner translations of RFT into a structured curriculum, and protocols are now embedded in PEAK Equivalence and PEAK Transformation modules.
RFT in Practice: Language Acquisition for Autistic Learners
This is where most BCBAs first meet RFT, even if nobody used the acronym at the time. Three programs are worth knowing.
PEAK Relational Training System
The Promoting the Emergence of Advanced Knowledge (PEAK) system, developed by Mark Dixon, is the most widely adopted RFT-informed curriculum in ABA. It has four modules: Direct Training, Generalization, Equivalence, and Transformation. The first two extend Skinner’s verbal operants. The last two operationalize derived relational responding and transformation of stimulus functions into discrete teaching programs. PEAK is assessment-driven, has a published item bank, and has accumulated a respectable evidence base for use with autistic learners and learners with developmental disabilities.
T-FT (Training in Frame Theory) and equivalence-based instruction
T-FT and the broader category of equivalence-based instruction (EBI) use the math of derived relations to make teaching efficient. Train the minimum number of relations the learner needs and let derivation do the rest. For a child learning sight words, you don’t have to train every word-picture-spoken triad in all six possible directions. Train two, get six free, if the equivalence class forms. EBI works for academic content too — fractions, foreign-language vocabulary, anatomy terms.
Equivalence-based instruction as a planning lens
Even if you never adopt PEAK formally, the RFT lens changes how you plan. When you write a goal, ask: which frame does this require? If the goal is “match printed word to picture,” you’re targeting coordination. If the goal is “identify which of three is biggest,” you’re targeting comparison. If the goal is “predict how your friend feels when you take their toy,” you’re targeting deictic and causal frames simultaneously. Naming the frame keeps you honest about whether your teaching procedure actually trains the relation you say it does.
RFT and ACT: How Theory Became Therapy
Acceptance and Commitment Therapy (ACT) is RFT applied to human suffering. Steven Hayes did not invent ACT and then go looking for a theory. He and his collaborators developed RFT and ACT in parallel from the late 1980s onward, and the two were always supposed to fit together.
The bridge is this: humans suffer because of how words and thoughts function, and words and thoughts function the way they do because of relational frames. A person who has framed themselves as “broken” sits inside a relational network where “broken” is in coordination with “me,” “me” is in coordination with the self-as-content, “broken” is in comparison with “fixed” people, and the transformation of stimulus functions makes ordinary social situations feel like evidence of brokenness. You can’t argue someone out of that network because the argument enters the same network and gets framed accordingly.
ACT’s six processes — defusion, acceptance, contact with the present moment, self-as-context, values, committed action — are each designed to alter relational responding in specific ways. Defusion weakens the literal coordination between a thought and its referent (“I’m having the thought that I’m broken” is in a different frame than “I’m broken”). Self-as-context uses hierarchical and deictic frames to build a perspective from which thoughts can be observed rather than identified with. Values work uses temporal and causal frames to link present action to a chosen future. Whether or not a clinician needs to explain any of that to a client (usually not), the procedures derive from RFT principles.
The IRAP (Implicit Relational Assessment Procedure)
The IRAP, developed by Dermot Barnes-Holmes and colleagues, is RFT’s measurement tool. It presents stimulus pairs and contextual cues, then asks the participant to respond consistently or inconsistently with a hypothesized relation under time pressure. The difference in response latency between consistent and inconsistent trials indexes the strength of the relational response.
The IRAP is used in research on implicit attitudes (race, gender, body image), clinical populations (depression, anxiety, OCD self-content), and as a process measure in ACT studies. It is not a clinical assessment most BCBAs will administer day to day. It’s worth knowing it exists because it grounds RFT empirically — derived relations can be measured, manipulated, and shown to predict behavior outside the lab.
Common Misunderstandings About RFT
“RFT replaces Skinner.” No. RFT extends Skinner. Mands, tacts, and echoics still exist and still need to be taught. RFT explains what happens after a learner has a critical mass of those operants and starts generating relations that weren’t trained.
“RFT is just stimulus equivalence with new vocabulary.” Equivalence is one frame (coordination) in the RFT family. RFT also covers opposition, distinction, comparison, hierarchical, temporal, causal, and deictic frames, each with its own contextual cues and clinical applications.
“You need to teach a kid RFT before they can talk.” You don’t teach RFT. You teach within frames. The frames themselves emerge from a learner’s verbal history. PEAK and similar programs structure that history.
“ACT is mindfulness for ABA people.” ACT borrows mindfulness procedures but the underlying model is relational framing, not contemplative tradition. The procedures look similar; the mechanism is different.
“If a learner shows derived relations, they have language.” Derived relations are part of language, but language also requires social conventions, syntactic structure, and pragmatic use. RFT explains a critical piece, not the whole pie.
Where RFT Is Contested
It would be dishonest to present RFT as settled science. It isn’t. Three areas of legitimate debate:
Mechanism. Whether derived relational responding is best explained as a generalized operant (the RFT account), as a basic stimulus phenomenon (Sidman’s equivalence account), or as something else, is still argued in the literature. Both camps generate empirical work and the data don’t cleanly favor one model in every paradigm.
Adoption within behavior analysis. Many practicing BCBAs do solid work without ever invoking RFT. Some senior behavior analysts have argued that RFT introduces mentalistic-sounding constructs through the back door. The RFT response is that “relating” is a behavior, not a mental state, but the debate continues.
Clinical efficacy of RFT-derived procedures vs. standard ABA. PEAK and EBI have evidence behind them, but head-to-head trials against equally well-implemented standard verbal behavior programs are not abundant. The pragmatic case for RFT-informed teaching is strongest when standard procedures have hit a wall — generalization isn’t happening, derived performances aren’t emerging, perspective-taking is stuck.
None of this means you should ignore RFT. It means you should hold it the way you should hold any framework: as a useful lens, evaluated against client outcomes, not as a creed.
A Worked Example: Teaching Derived Naming in a Clinical Session
Imagine a five-year-old client, Mia, who has solid tacts for common objects but doesn’t reliably show derived relations. You want to build a small coordination-frame network around three new items: a picture of a kiwi (A), the spoken word “kiwi” (B), and the printed word KIWI (C).
Step 1 — Direct training. Teach A→B: show the picture, prompt the echoic “kiwi,” reinforce. Run to mastery. Then teach B→C: say “kiwi,” prompt selection of the printed word from a field of three, reinforce. Run to mastery.
Step 2 — Probe derived relations. Without further training, test:
- B→A: say “kiwi,” ask Mia to select the picture.
- C→B: show the printed word, ask her to tact it.
- A→C: show the picture, ask her to select the printed word.
- C→A: show the printed word, ask her to select the picture.
Step 3 — Interpret. If all four derived relations emerge on the first probe, Mia is showing combinatorial entailment within a coordination frame. The equivalence class is intact. You’ve taught two relations and gotten six. If only one or two emerge, you have data: she may need more exemplars to abstract the “same as” frame, or she may benefit from multiple-exemplar training across many three-stimulus sets before the frame generalizes.
Step 4 — Transformation probe. Pair the picture of the kiwi with a small taste test. If Mia likes the taste, present the printed word KIWI later in the day with no picture and observe whether she approaches, requests, or shows other appetitive behavior toward it. If she does, you’ve demonstrated transformation of stimulus functions — the function transferred from picture to printed word via the coordination frame.
That’s RFT in a session. No mysticism, no jargon at the client level, just a structured way to plan teaching and interpret what shows up in probes.
Frequently Asked Questions
Is RFT evidence-based?
RFT has hundreds of basic-research studies supporting its core claims about derived relations, mutual and combinatorial entailment, and transformation of stimulus functions. RFT-derived clinical procedures (PEAK, EBI, ACT) have varying evidence bases — ACT is well-supported across multiple populations, PEAK has growing evidence, and EBI has decades of educational research. The theory and the procedures are in different evidentiary positions, and it’s fair to evaluate them separately.
Do I need to know RFT to be a competent BCBA?
No. You can deliver effective behavior analytic services without invoking RFT at all. You will, however, run into clinical situations — stalled language programs, missing perspective-taking, clients with rigid verbal rules — where RFT gives you a more precise lens than standard verbal behavior alone. Treat it as a useful tool, not a credential.
How is RFT different from stimulus equivalence?
Stimulus equivalence is a specific phenomenon: the spontaneous emergence of symmetrical, reflexive, and transitive relations after conditional discrimination training. RFT treats equivalence as one frame (coordination) and adds opposition, distinction, comparison, hierarchical, temporal, causal, and deictic frames as additional generalized operants. RFT also adds transformation of stimulus functions, which equivalence theory does not centrally address.
Is ACT just RFT applied to clinical problems?
Largely yes. ACT’s six processes map onto specific predictions RFT makes about how relational networks influence behavior. That said, ACT pulls clinical procedures from multiple traditions (mindfulness, exposure, behavioral activation, values clarification) and combines them in ways that aren’t strictly derivable from RFT alone. Hayes himself has been clear that ACT can be practiced without explicit RFT knowledge.
Can RFT predict who responds to ACT?
Process research using the IRAP and other measures has shown that changes in relational responding during ACT correlate with clinical outcomes. We can’t yet take a baseline IRAP and confidently predict individual responder status, but the mechanism work is more advanced than for most therapies.
Is RFT in the BCBA task list?
The current BACB task list covers verbal behavior, stimulus equivalence, and rule-governed behavior, all of which intersect with RFT, but RFT is not named as a required content area. You can pass the exam without a dedicated RFT course. Many supervisors still include it because it sharpens clinical reasoning around language and rule-following.
How do I learn RFT well?
Three steps. First, read the Törneke book listed below — it’s the most readable introduction. Second, watch CEU presentations from clinicians who use RFT in session, because seeing it applied makes the abstract concepts stick. Third, run a small EBI or PEAK program with a real client and pay close attention to what does and doesn’t derive. Theory clicks when probes don’t behave the way you predicted.
Can RBTs use RFT-based teaching?
RBTs implement programs that BCBAs design, and RFT-informed programs are no exception. PEAK protocols are written to be implemented by paraprofessionals with BCBA supervision. RBTs don’t need to lecture on combinatorial entailment, but they do need clean prompting, clear contextual cues, and accurate data collection on derived probes.
Does RFT work with non-vocal learners?
Yes, and this is one of its more interesting applications. Derived relations have been demonstrated in learners using AAC, sign, and PECS. The relational responding is in the selecting and matching, not in the vocalization. EBI and PEAK both have protocols adapted for non-vocal communicators.
What’s the fastest red flag that a program is misusing RFT?
Vague claims that “RFT will fix” perspective-taking, anxiety, or social skills without specifying which frames are being targeted, which contextual cues are being trained, and what derived performances will be probed. RFT-informed practice is concrete and measurable. If a program can’t tell you what derivation it expects to see, it’s branding, not RFT.
Where to Learn More
If this guide has you wanting more, here’s where to go next.
- Free CEUs on RFT-adjacent topics — BBC runs a rotating library of free, BACB-eligible CEUs at behavioristbookclub.com/free-aba-ceus. Verbal behavior, ACT for BCBAs, and equivalence-based instruction show up regularly.
- Paid + free CEU library — for a fuller catalog of clinically relevant CEUs, including deeper RFT and ACT content, see the BBC CEU library.
- Niklas Törneke, Learning RFT: An Introduction to Relational Frame Theory and Its Clinical Application — the most accessible book-length introduction. Start here.
- Hayes, Barnes-Holmes, and Roche (Eds.), Relational Frame Theory: A Post-Skinnerian Account of Human Language and Cognition — the canonical academic text. Dense but foundational.
- PEAK Relational Training System — Mark Dixon’s curriculum. The PEAK Direct Training, Generalization, Equivalence, and Transformation modules are how most BCBAs operationalize RFT in clinic.
- Hayes, Strosahl, and Wilson, Acceptance and Commitment Therapy (2nd ed.) — the ACT side of the lineage. Read it after Törneke so you can see the RFT bones underneath the clinical protocol.
RFT is one of those topics where the first read leaves you suspicious and the third read leaves you converted. It rewards patience. And it pays off in the room, when a kid finally derives a relation you didn’t train, or a client finally notices that a thought is a thought and not a fact. That’s the moment the theory stops feeling academic and starts feeling like the most practical thing in your toolkit.