Professional Documents
Culture Documents
Stimulus equivalence
Material selected from Cooper J.O, Heron T.E, Heward W.L. Applied behavior analysis (3nd
ed.) 2020
When successful performances are obtained on probe trials of all three types—reflexivity,
symmetry, and transitivity (or combined tests for equivalence)—in the absence of reinforcement,
the defining criteria for equivalence-class formation are satisfied.
Each stimulus from a given class functioned identically, whether as a sample or as a
comparison stimulus, in relation to any other member of the same class. It is exactly this
emergent interchangeability that allows us to interpret the stimuli within each class as symbolic
of one another (e.g., a symbolic relation was demonstrated between the spoken word, “car,” the
written word, [CAR], and the picture of the car). What’s more, this interchangeability was the
result of a structured teaching approach involving only a small subset of the total number of
relations generated.
Class formation—typically measured as the percentage of probe trials on which a comparison
selection is consistent with equivalence, for each defining probe type. Such measures are often
compared in a pretest, train, posttest design, or in a multiple baseline across stimulus sets
experimental design.
Translational research represents a reciprocal and dynamic process in which the laboratory and
the field inform and inspire each other, leading to increased effectiveness in both domains, and
providing for the strength of behavior analysis as a discipline.
Class merger—a combining of independent equivalence classes such that a larger class is
established. This is typically accomplished by teaching a new relation between one member of
each of the classes to be combined. Class merger is verified by emergent relations between the
other stimuli from the formerly independent classes. Class merger describes an outcome in which
stimuli from the originally independent classes A1B1C1 and D1E1F1 are matched to each other,
such that a single six-member class (A1B1C1D1E1F1) emerges.
Transfer of function—the finding that teaching a new function for one member of an
established equivalence class results in the same function holding for all members of the class.
Contextual control—stimulus control requiring three levels of antecedent stimuli, such that the
functions of the stimuli in a conditional discrimination vary depending on the context.
Contextual control training requires a five-term contingency. It allows for the same stimuli to be
members of more than one equivalence class, depending on the context
Selecting a given comparison stimulus (e.g., B1) is reinforced in the presence of a particular
sample (e.g., A1) in one condition (i.e., Context 1), while selecting a different comparison
stimulus (e.g., B2) in the presence of the same sample (e.g., A1) is reinforced in a different
condition (i.e., Context 2).
2.observing responses-for example, touching or naming each comparison in the array prior to
the selection response should increase contact with each of the choices, and thereby increase the
likelihood of control by the correct comparison
3.instructions- varied from simple (e.g., “Touch one”; Lynch & Green, 1991) to more detailed
trial-type arrangements and sequences- ensure an equal opportunity for reinforcement with
each comparison. Imbalance here could create a bias toward selecting one (or more)
comparison(s) over the others. Further, the position of the comparison stimuli should be
counterbalanced across trials, such that each stimulus appears in each possible position on an
equal number of trials. Imbalance here could result in a conditional relation between the sample
stimulus and a particular position, rather than a particular comparison. For example, young
children may lose interest in the task after a relatively limited number of trials, so blocks ranging
from 9 to 27 trials are common, while college students will often work for periods of 1 to 3hours
5.training structure-the teaching arrangement (nodal stimulus set or, more simply, the node,
must be held in common across a minimum of two conditional discriminations to provide a basis
for all equivalence properties. One-to-many training, in that one set of sample stimuli (e.g., the
A stimuli in this case) is trained with multiple sets of comparison stimuli (e.g., B and C, in this
example). A synonymous term for this training structure is sample-as-node training, which
emphasizes that the A stimulus set serves a sample function in both of the trained conditional
discriminations and thus provides the basis for emergent relations between stimuli from the
different conditional discriminations (e.g., between B and C stimuli, in this case). A second
popular training protocol is the many-to-one or comparison-as-node structure. In this
arrangement, multiple sets of sample stimuli are trained with a single set of comparisons, and the
comparison stimuli thus serve as node. For example, AB and CB conditional discriminations
could be trained such that the B comparison stimulus set is held in common. A third structure is
known as a linear series training. Here, the comparison stimuli from one conditional
discrimination serve as sample stimuli in the next; for example, AB and BC relations might be
taught.
6. mastery (accuracy) criteria 88% with no more than one error on any given trial type, for two
consecutive sessions.
7.and consequence delivery. p. For example, the steps might include having reinforcers
available on 100%, then 75%, and then 50% of the trials
Matching-to-sample testing protocols
Composition of probe trial Two general methods are used for testing: massed testing and
interspersing probe trials with baseline-trial types. In the first method, the trial block is
composed of probe trials only. In the second, the block includes probe trials as well as each of
the baseline training trial types presented in a semirandomized order, often with no more than
one or two probe trials presented in succession.
Scheduling Consequences During Testing extinction conditions are in effect for all trial types—
baselines as well as probes
Test Order common practice in EBI is to present only one probe type (i.e., reflexivity, symmetry,
transitivity and/or combined equivalence) per test-trial block.
For programs of EBI with young children or individuals with intellectual disabilities, where the
generation of equivalence classes is of primary interest, a simple-to-complex testing protocol
has much to recommend it In this arrangement, probe types are introduced sequentially,
beginning with symmetry, followed by transitivity (if relevant), and then combined tests for
equivalence. Similarly, probe types requiring fewer nodes are presented prior to those requiring
more nodes.
Judging Class Formation- high levels of classconsistent responding (e.g., 90% or greater).
When the expected emergent performances are not demonstrated immediately, several different
strategies have proven helpful: re-exposure to the training blocks, training or testing
modifications that target particular forms of competing stimulus control. For example- a
position- the presentation format could be redesigned so that no stimulus appears in that position
on any trial.
Naming theory and relational frame theory (RFT) provide differing accounts of the basis for
novel relational responding. Both emphasize that novel responding is due to a special learning
history, rather than a direct outcome of a reinforcement contingency (as in Sidman’s view).
Naming has been suggested (Horne & Lowe, 1996) as a technical term for a higher-order
operant in which an individual engages in bidirectional speaker and listener behavior with
respect to a given object or event. For certain participant populations, teaching critical
components of the naming unit (i.e., listener behavior, tacting) can be effective in generating
repertoires indicative of symbolic relations, classes, and categories
In relational frame theory (RFT), framing is viewed as the key higher-order operant, and it is
characterized as responding to one stimulus in terms of another (Hayes et al., 2001). RFT
emphasizes that we learn many different types of relations in our lives (e.g., larger than, opposite
of, different from), each of which can be considered a frame in that any pair of stimuli can
function within it.
RFT introduces generic terms
-mutual entailment (e.g., if A is larger than B, then B is smaller than A; in equivalence
relations, symmetry would be an example of mutual entailment);
-combinatorial entailment (e.g., if A is larger than B, and B is larger than C, then A is larger
than C, and C is smaller than A; in equivalence relations, transitivity would be an example of
combinatorial entailment);
-transformation of function (e.g., if A is larger than B, and B is good, then A would be better;
in equivalence relations, transfer of exactly the same function from one stimulus to another in its
class would be an example).
Equivalence -known as the frame of coordination in RFT terminology.
Multiple-exemplar training is viewed as the basis for learning to frame.