How to test RFT

How to test RFT

HOW TO TEST RFT

I occasionally hear the old chestnut that "RFT is not really testable."

It makes me mad. 

Over two decades ago my students and I laid out a fairly well crafted list of ways you can test RFT.

I've put the in the page numbers of the quotes below for a reason: so you can cite specific predictions if you want even if you do not have access to this book.

Many of these predictions are now known to be true. And not one piece of disconfirmatory evidence has yet emerged, so far as I know.

I would invite RFTers who are publishing new pieces to occasionally remember our history. If you agree with me that these were indeed reasonable stakes to pound in the ground 20 years ago, then every once in a while it would be good remind readers that the theory was laid out in testable ways from the beginning. There are dozens of such predictions in the 2001 purple book but even before that, we were putting our ideas on the line and asking for help in showing where and how the theory was wrong. In the history of science all theories are wrong -- in their details at least -- given enough time and effort to test them. That surely includes RFT. I'm not a falsificationist but risky tests are pragmatically useful because as we learn more about the contextual conditions under which knowledge claims, we advance the precision, scope, and depth of our analyses.

Anyway, the summary below should be of use.

- Steve Hayes

*******************************

Hayes S. C., Gifford, E. V. & Ruckstuhl, Jr., L. E. (1996) Relational frame theory and executive function.  Chapter in G. R. Lyon & N. A. Krasnegor (Eds.), Attention, memory and executive function (pp. 279-305). Baltimore: Brookes.

From p. 298:

TESTS OF THE THEORY

A theory of executive function based on relational frames has two somewhat distinct components that can be tested: whether relational frame theory is a worthwhile account of human verbal behavior and verbal regulation, and whether executive function can be usefully thought of in these terms. 

Relational frame theory argues that relational frames are learned and are not primitive psychological functions. We already mentioned five testable components to this claim: They should show clear developmental trends, they should be flexible, they should be under both antecedent and consequential control, and improved relational abilities should emerge from deliberate training. Supportive data exist in each these five areas, but much more remains to be done.

The last point is particularly in accord with the pragmatic assumptions of behavior analysis: The best way to test a theory or device is to see if it can lead to more effective treatment (Hayes, Nelson, & Jarrett, 1987). Training could occur with populations who have disabilities in rule generation, rule understanding, and rule following, such as children with attention deficits or hyperactivity, youth who are antisocial, and those with mental retardation. Typical youth could be

From p. 299:

given procedures designed to modify existing repertoires and accelerate their healthy development.

These five testable components of the claim that relational frames are learned also suggest ways that this key aspect of relational frame theory can be disproved. If derived stimulus relations are present in whole cloth in neonates, or emerge without training in nonhumans, the theory is disproven. If new, more subtle, or more complex stimulus relations cannot be taught to children and brought to bear on arbitrary events in a fashion envisioned by the theory, it is disproven. The theory argues that a wide variety of relations can be trained and that derived stimulus functions will be transformed by these underlying relations. The first point has some empirical support, but not yet the latter. If complex relations are merely a by-product of equivalence and nonequivalence, and if derived functions are merely transferred, not transformed, when relations such as oppositeness pertain to two stimuli, the theory is disproven. A key idea is that relational frames are a defining feature of human verbal behavior. If the behavioral functions of verbal events (e.g., self-awareness, construction of long-term futures and a resultant reduction in impulsivity, complex problem solving) do not emerge in children in a fashion that parallels developing relational abilities, the theory is disproven. If complex relational frames can be developed in nonhumans, without also seeing some of the effects produced in humans by verbal behavior, the theory is disproven.

If relational frame theory is correct, children should show increased abilities in verbal regulation as they learn to apply more complex relational frames to events (e.g., if-then, comparatives). The theory suggests that the key aspects of training are 1) increases in the number of available relational frames, 2) increases in combinatorial abilities and the resultant complexity of the derived relational networks, 3) greater sensitivity and subtlety in the contextual control of relational frames and resulting increases in both their arbitrary applicability and appropriate regulation by physical context, 4) increased ability to transform stimulus functions through derived stimulus relations and greater sensitivity and subtlety in the contextual control over this process of transformation, 5) greater ability to relate networks of relations, and 6) greater ability to alter the functions of the previously nonverbal world by including aspects of this world in relational networks. As these performances increase in children, we should see increases in self-control, reasoning, and problem solving-if not, the theory is disproven.

According to the theory, following verbal rules is a product of the ability both to apply if-then frames to events and to transform the functions of verbally constructed consequences and of experience with contingencies that support specific types of rule following, such as pliance, tracking, and augmenting. Both contentions are clearly testable. Our line of thinking also suggests that pliance usually helps establish tracking, which helps establish augmenting. It suggests that moral development and other complex forms of rule governance normally emerge in that sequence and may need to be trained in that sequence. It also suggests novel ways that deficits in rule following may occur, such as mismatches of 

From p. 300:  

types of rules and rule following (e.g., the tendency for persons with some histories to treat descriptions as demands and thus to show pliance or counterpliance instead of tracking in these situations).

Testing these effects of verbal rules will be difficult, but behavior analysis offers the field at large not just a theoretical approach, but also a set of methods that are highly precise and well-developed. These include refined methods of arbitrary matching-to-sample and methods for testing the effects of rules on sensitivity to changes in environmental demands. Behavior analysts have also argued for and used refined "talk aloud" methods for detecting the participation of verbal rules in problem solving (Hayes, 1986; Wulfert, Dougher, & Greenway, 1991). 
 

Steven Hayes