Professional Documents
Culture Documents
1 (Spring)
of stimuli that make them verbal FASs examine how one may attempt to sepa-
are not clear. Elsewhere it has been ar- rate their respective effects in experi-
gued that such stimuli most likely possess ments on rule-governed behavior.
certain formal properties (Schlinger,
1990; Schlinger & Blakely, 1987). Simply A Function-Altering Versus
stated, it was suggested that such stimuli Discrimination Interpretation
must specify the events to be related; in of "Rules"
other words, to borrow a term from Skin-
ner (1969), they must be contingency- Suggesting that some verbal stimuli
specifying stimuli (CSSs). In other words, have function-altering effects is one thing;
a function-altering CSS must name at demonstrating those effects is another.
least two events (e.g., two stimuli, or an Such a demonstration is especially diffi-
antecedent stimulus and a response, or a cult when attempting to separate dis-
response and a consequence, etc.). Of crimmnative from function-altering ef-
course, three (or more) events may also fects. An even more difficult task is to try
be included, typically a stimulus, a re- to convince behavior analysts that a dis-
sponse, and a consequence. When for- crimination theory of rules (e.g., Cerutti,
mally defined, most instances of verbal 1989) may be inadequate to explain all
statements that produce function-alter- of the effects of verbal stimuli. Consider
ing effects may be said to be contingency the following example: "When you come
specifying only in the sense that the for- to a red light put on the brakes by press-
mal verbal unit contains at least two ing the pedal on the left." Now, if such
terms. However, not all verbal FASs need a statement controls the appropriate be-
contain two or more terms. In sophisti- havior, classifying it as an SD seems nat-
cated listeners it is common for a single ural. According to Malott (1989), an SD
word to have function-altering effects. For is a stimulus in the presence of which a
example, if someone is about to eat some particular behavioral contingency is more
food and someone else says, "poison," likely to be operative. However, it is dif-
this alters the evocative function of the ficult to find in this example a behavioral
sight of the food in the sense that it de- contingency occurring in the presence of
creases the momentary probability of the a stimulus that we can then call an SD. If
behavior of putting it in the mouth. Some we use Michael's (1983) definition of the
might suggest that there is an implicit SD and include an evocative (or occa-
CSS in the word "poison" (e.g., "If you sioning) requirement, then there would
eat this food you will get very sick"). have to be a momentary increase in the
However, such a suggestion begs the probability of the behavior controlled by
question concerning the function of the the rule (Michael, 1983). Although we
verbal stimulus. Thus, irrespective of the might be able to make a case for the red
form of the verbal stimulus, if it is func- light as the SD for pressing the pedal,
tion altering then we may speak of it as making a case for the statement as the SD
a rule. is more difficult.
Verbal FASs can alter several behav- Separating discriminative from func-
ioral functions, including (a) the evoca- tion-altering effects of rules becomes even
tive functions of stimuli that mimic the trickier when a CSS is stated and alters
evocative effects of discriminative and the function of some event that is already
motivational events, (b) the evocative present, such that the relevant behavior
functions of stimuli that mimic respon- immediately occurs. In these situations
dent conditional stimuli, and (c) the re- the temporal proximity between the CSS
inforcing or punishing functions of stim- and the behavior it specifies makes it ap-
uli that mimic the same effects that result pear that the CSS evoked the behavior,
from nonverbal procedures. thus strengthening an SD interpretation.
Given this briefdescription ofdiscrim- Consider the following instruction, taken
inative and function-altering effects of from Ayllon and Azrin's (1964) study
verbal stimuli, we may now move on to with mental patients, which compared
DISCRIMINATIVE AND FUNCTION-ALTERING EFFECTS 13
reinforcement with instructions for pick- GSR and heart rate as if it had been cor-
ing up eating utensils: "Please pick up related with the shock in a respondent
your knife, fork, and spoon, and you have conditioning procedure. The question is
a choice of extra milk, coffee, cigarettes, this: How did the bell acquire such evoc-
or candy." Does this statement function ative functions? The answer seems clear-
as a discriminative or a function-altering ly to implicate the CSS, but not as an SD.
CSS? In other words, does the CSS evoke The human operant literature is replete
the relevant behavior only because of a with examples of CSSs that seem to en-
history, even generalized, of discrimi- dow various stimuli (e.g., lights and
nation training, or does the CSS alter the sounds) with potential reinforcing or
function of the sight of the utensils? In punishing properties. For example, in a
this case, the utensils were presumably study with young children using a re-
within sight of the patient at the time of peated-acquisition design, Vaughan
the instruction. So, was picking-up be- (1985) stated, among other things, that,
havior evoked by the combination of the "Sometimes you will hear a nice sound
statement and the sight of the utensils (as for pressing a button, but at other times
a conditional SD), or did the CSS alter you will hear a loud buzzer telling you
the evocative (SD-like) function of the that you have pushed the wrong button"
utensils? Obviously, it is extremely dif- (p. 177). In another example, Danforth,
ficult to tell. Of course, there is always Chase, Dolan, and Joyce (1990) stated to
the possibility that the CSS may have college students who served as subjects,
both effects simultaneously. In fact, it is "Each time you make a correct response
only when there is greater temporal sep- a beeper will sound" (p. 99). Finally, in
aration between a CSS and the specified a study by Matthews, Shimoff, Catania,
behavior that we may more clearly point and Sagvolden (1977), college students
out its function-altering effects. This does read the following "instructions": "Each
not mean that function-altering effects point is worth 0.1 ¢. For example, if you
occur only after long temporal delays. earn 2,000 points, you will be paid $2.00"
Some behavior analysts would still ap- (p. 455). In each of these cases, different
peal to a discrimination theory to explain stimulus changes-points and sounds in
such relationships, irrespective of the the Vaughan (1985) experiment, sounds
temporal delay (e.g., Cerutti, 1989). in the Danforth et al. (1990) experiment,
However, as was recently pointed out, and points in the Matthews et al. (1977)
there is also some disagreement among study-were endowed by the CSSs with
behavior analysts concerning the neces- reinforcing (or punishing) properties.
sary and sufficient defining properties of These changes in stimulus functions could
SDS (Schlinger et al., 1991). easily be tested by comparing perfor-
Fortunately, the best evidence against mance on a simple task before and after
a discrimination interpretation of FASs the stated CSS. It is difficult to imagine
comes from examples of CSSs that affect how a discrimination interpretation could
behavior but that don't seem explainable explain these effects.
in terms of a discrimination interpreta- Following from the logic that CSSs can
tion. As stated previously, verbal stimuli alter respondent-like evocative and re-
can alter respondent-like stimulus func- inforcing or punishing functions of stim-
tions, including those that endow pre- uli, we are led to conclude that SD-like
viously neutral stimuli with reinforcing evocative functions can also be altered.
or punishing effects. In Verbal Behavior, For example, Vaughan (1985) instructed
Skinner (1957) offered the example, her children, "Do not begin pushing but-
"When you hear the bell you will feel a tons until the panel lights come on" (p.
shock." Now, when the bell rings how do 178). In addition, some of the subjects
we account for increases in the subject's in the Matthews et al. (1977) study read,
GSR and heart rate? Certainly not in "Whenever the RED LIGHTS beside the
terms of operant stimulus control by the counter are on, each press on the RED
statement. In this example the bell evokes BUTTON will add one point to your to-
14 HENRY D. SCHLINGER, JR.
tal" (p. 455). Vaughan's instruction spec- lott (1989) distinguishes between contin-
ifies stimuli and responses (panel lights gencies that are direct-acting and those
and button presses), whereas the instruc- that are not direct-acting. Direct-acting
tion of Matthews et al. specifies a three- contingencies involve effective behavior-
term contingency -red lights (and the red al outcomes (consequences) for the causal
button), button presses, and points. In response, whereas indirect-acting contin-
both cases, however, appropriate re- gencies involve outcomes that are not ef-
sponding was brought under the control fective for the causal response; in other
of the relevant stimulus as a result of the words the behavioral outcome is too de-
CSS or, said another way, the CSS en- layed, too improbable, or too small to be
dowed the relevant stimuli with evoca- effective. For example, the improved
tive (SD-like) functions. Of course, some health resulting from a low-fat diet is too
might still argue that the statements are delayed to reinforce the behavior of eat-
SDs for the behavior. However, there are ing healthy food. Obviously, direct-act-
more straightforward examples of dis- ing contingencies control behavior
criminative (instructional) control. For through known laws ofconditioning, but
example, in the Matthews et al. (1977) what about indirect-acting contingen-
study, the last line of the instruction card cies? They are defined on the basis of
read, "Put on the headphones now. . . ," structural rather than functional char-
which presumably evoked the appropri- acteristics; that is, they involve outcomes
ate behavior as a (conditional) SD (or pos- that do not function as effective behav-
sibly as an EO). The next part of the in- ioral consequences. So what behavioral
struction, however, read ". . . and do not function do they have? It is difficult to
remove them until the end of the ses- imagine, because indirect-acting contin-
sion" (p. 455). This CSS controls behav- gencies by definition do not act. Never-
ior by altering the stimuli that signal the theless, Malott (1989) distinguishes be-
end of the session such that they evoke tween rules that specify direct-acting
the behavior of removing the head- versus indirect-acting contingencies and,
phones, and is thus function altering. moreover, between rules that are com-
plete versus incomplete. But again, these
are structural and not functional distinc-
SOME IMPLICATIONS OF tions. And what about the function of
FAILING TO DISTINGUISH rules? Of course, indirect-acting (or "re-
BETWEEN DISCRIMINATIVE AND mote," see Michael, 1986) contingencies
FUNCTION-ALTERING CONTROL are interesting precisely because at times
Now that we have differentiated dis- they do seem to affect behavior. A dis-
criminative from function-altering ef- crimination (or motivation) interpreta-
fects, further discussion is possible of the tion, however, would seem to fall short
implications of failing to make the dis- in explaining how such long spatio-tem-
tinction, especially in research involving poral gaps can be mediated. Of course,
the control of behavior by verbal events. we could always bridge the gap with cog-
nitive -mediational-analyses. The
Semantic Implications: The Language present paper argues that perhaps one,
of Rule-Governed Behavior indeed the only, way that such indirect-
acting "rules" might affect behavior is by
The first implication of failing to dis- altering the behavioral functions of the
tinguish between discriminative and events specified by the rules.
function-altering effects of verbal stimuli A second semantic problem with be-
concerns how behavior analysts talk havior analysts' treatment of stimuli
about verbal events. First, in many in- called rules is that in many instances when
stances behavior analysts have fallen into behavior analysts have taken a functional
the (cognitive) trap of talking about rules approach to rule-governed behavior, they
as structural entities. For instance, Ma- have confused discriminative and func-
DISCRIMINATIVE AND FUNCTION-ALTERING EFFECTS 15
tion-altering effects, usually by failing to hence, they open the door to cognitive
correctly identify function-altering ef- interpretations.
fects. The result, as might be expected, is
that the language of discrimination (or Methodological Implications:
motivation) is used to describe function- The Function(s) of Instructions
altering processes. For example, behav-
ior analysts have defined rules variously The methodological implications of
as "antecedent verbal stimuli" (e.g., Ca- failing to distinguish between discrimi-
tania et al., 1990; Hayes, 1986)-a struc- native and function-altering effects of
tural definition-or as a "verbal descrip- verbal stimuli are related to the semantic
tion of a contingency" (Braam & Malott, implications. In other words, the ways in
1990) -another structural definition. And which behavior analysts talk about the
although the function of these antecedent verbal stimuli they speak ofas rules have
verbal stimuli that describe contingen- determined the types of research into their
cies is rarely specified, it is frequently effects. As stated earlier, behavior ana-
implied as being discriminative. For ex- lysts typically interpret rules (or instruc-
ample, Catania et al. (1990) describe rule- tions) as SDS. Consider the term instruc-
governed performances as "occasioned," tion. For several researchers, an
the same term used to refer to control by instructional stimulus is one that is cor-
SDS. Presumably rules are SDS that "oc- related with reinforcement for correct re-
casion" the performances. Braam and sponding or, in other words, a discrimi-
Malott (1990) define rule-governed be- native stimulus. For example, in
havior as behavior under the control of repeated-acquisition experiments, Boren
a rule. But they define a rule as a "verbal and Devine (1968), Vaughan (1985), and
description of a contingency." What kind Danforth et al. (1990) all programmed
of control is implied? Interestingly, they stimuli that accompanied each correct re-
suggest that rules function as establishing sponse in the chain. In the studies by
operations. In both cases, however, rule Boren and Devine (1968) and Vaughan
control would have to be seen as evoc- (1 985), a light appeared above the correct
ative and not as function altering; that is, response, and for Danforth et al. (1990)
the rule evokes the relevant behavior as either a small dot or the phrase, "This
a part of a (direct-acting) contingency of one is correct," accompanied correct re-
reinforcement like any other SD or EO. sponses. Although the effects of these
Behavior analysts should not make the "instructions" on behavior are interest-
mistake of assuming that there are essen- ing, if the effects are only discriminative,
tial properties of rules that distinguish then it is doubtful that we learned much
them from other stimuli (see Palmer & that is new about any unique functions
Donahoe, 1992). The position of the of "instructions" or, according to the
present paper is that verbal stimuli must present interpretation, about rule-gov-
be classified according to their function, erned behavior.
and if some verbal stimuli are no differ- In much of the human operant re-
ent than their nonverbal counterparts, search, the specific function of instruc-
they should not be distinguished with a tions has not been considered. In fact, in
special term (see also Blakely & Schlin- many studies the instructions are defined
ger, 1987; Schlinger, 1990). more by their structural properties, such
Finally, another semantic byproduct of as whether the instructions specify dead-
a discrimination interpretation of rules is lines or immediate or delayed conse-
the use oflay-cognitive terminology such quences (e.g., Braam & Malott, 1990).
as "rule following," "following instruc- Their function is usually considered in a
tions," "rule compliance," or "obeying more global fashion in terms of whether
rules." These and other ways of describ- their presence has any effect on respond-
ing rule-governed behavior are organism ing. Sometimes the effect of interest is
based and not environment based and, whether the presence of instructions is
16 HENRY D. SCHLINGER, JR.
examined the effects of deadlines and de- havior controlled by it would have oc-
layed consequences on rule compliance curred fairly quickly. That is, it would
in preschool children. In their study the have been evoked by the instruction (Mi-
following conditions were compared: (a) chael, 1982). Indeed, in a study that rep-
no deadline and no reinforcer, (b) im- licated and extended Braam and Malott's
mediate deadline and immediate rein- (1990) study, Mistr (1992) noted that
forcer, (c) immediate deadline and no re- when the opportunity to respond was im-
inforcer, (d) no deadline and delayed mediately available, most ofthe children
reinforcer, and (e) immediate deadline responded immediately regardless of the
and a delayed reinforcer. Each ofthe con- deadline. So, what is interesting about the
ditions was accompanied by a statement Braam and Malott study? The answer
from the experimenter; for example, in must be that the results are interesting if
the condition with immediate deadlines we get caught up in the structural inter-
and delayed reinforcers, the experiment- pretation of certain verbal stimuli, for
er said, "Here's a puzzle you can do. I example, whether "rules" specify direct
don't care if you do it or not. Here's the or indirect-acting contingencies.3 But are
rule: If you do the puzzle, now, one week these, as Braam and Malott suggest, re-
after you finish the puzzle, you can go to ally the "critical components" of rules?
the Magic Box" (p. 71). Braam and Ma- Based on the arguments in the present
lott found, not surprisingly, that the paper, we can suggest that verbal stimuli
conditions with immediate deadlines and that control behavior as SDS or EOs are
reinforcers produced the highest '"com- not that interesting for an experimental
pliance." However, they also found that or theoretical analysis of behavior and
immediate deadlines and delayed rein- that behavior analysts should attempt to
forcers resulted in fairly high compliance. discover whether there are other types of
Braam and Malott concluded that it is controlling relations that have yet to be
not the delay of the reinforcer that is crit- fully uncovered.
ically important in control by the "rules," One of the biggest problems of failing
but rather the presence or absence of a to distinguish between evocative and
specified deadline. Although Braam and function-altering effects of verbal stimuli
Malott referred throughout their article is that there has been no research into
to the instructions as SDs and S-deltas the provenance of function-altering ef-
(i.e., they used the language of discrim- fects. However, there are some data that
ination), they concluded that the "rules" suggest something about how such be-
controlled the relevant behavior as EOs. havioral relations might develop.
According to them, "the statement of the
rules established noncompliance with the STUDIES WITH IMPLICATIONS
rule as a learned aversive condition, so FOR THE DEVELOPMENT OF
compliance with the rule immediately, VERBAL FASs
and perhaps automatically, terminated An experiment on the relation between
or reduced this aversive condition" (p.
75). Perhaps they are right. Although they subjects' verbal behavior and the for-
attempted to control for the neutrality of mation ofequivalence classes by Wulfert,
the experimenter's instruction to the
children, the use of the word "now" may
have been the critical word (Mistr, 1992). 3 Braam and Malott (1990, p. 68) interestingly
It is likely that for many children the word defined rule-governed behavior as "behavior under
"now"l spoken by an adult has in the past the control of a rule," which is to be "contrasted
with behavior controlled directly by the contingen-
been correlated with other aversive events cy described by the rule." However, although they
that could be reduced by compliance. concluded that they were studying rule-governed
Braam and Malott (1 990) did not present behavior, they also concluded that the rules spec-
data indicating how quickly the children ifying deadlines functioned as EOs, thus "creating
a direct-acting contingency," which would mean
complied with the rule when they did, that they were not studying rule-governed behavior
but if the instruction did function as an at all, unless we define rules only in terms of their
EO, then we would expect that the be- structural, that is, syntactic, properties.
18 HENRY D. SCHLINGER, JR.
Dougher, and Greenway (1991) suggests sufficient to alter the evocative function
not only that equivalence-class forma- of stimuli in the symmetry or equiva-
tion depends upon the presence of verbal lence tests. Conversely, responding only
FASs, but also what the necessary verbal to the stimulus compound did not alter
constituents for such function-altering ef- the function of a novel stimulus to evoke
fects might be. In their first experiment, the correct match during an equivalence
Wulfert et al. had college students think test.
aloud as they responded on a matching- The subjects in the Wulfert et al. (I 99 1)
to-sample task. The experimenters then study were adults, therefore, it might be
correlated accuracy of responding on informative to examine the age at which
symmetry and equivalence tests with the such verbal responding normally ap-
type of verbalization for each subject. pears. An interesting experiment by Pou-
Verbalizations were classified according thas et al. (1990) seems to do just that.
to four different categories: (a) "relational Pouthas et al. were interested in whether
responding," which included references the duration of button pressing in chil-
to the relationship between two stimuli, dren aged 4.5, 7, or 11 years would con-
for example, "circle goes with the open form to scheduled contingencies of re-
triangle"; (b) "common physical fea- inforcement and, furthermore, whether
tures," which meant relating pairs of the children would correctly describe the
stimuli according to nonarbitrary fea- contingency and whether their behavior
tures, for example, "they look alike"; (c) would be more sensitive to their descrip-
"stimulus compounds," which meant that tions or the actual contingencies. For one
the subjects visually integrated the sam- group ofchildren the experimenters used
ple and comparison stimulus, for ex- posttrial probes, by asking after each tri-
ample, "together they look like a house"; al, "What did you have to do to get very
and (d) "other," which included all other good?" For a second group of children
responses including silence. The experi- the experimenters used postsession in-
menters found that those subjects who terviews comprised of similar questions.
did not exhibit equivalence rarely en- The results showed that the 1 -year-olds,
gaged in relational verbal responding, al- but not the 4.5- or 7-year-olds, were able
though they did name compound stimuli to correctly describe the "rule" for cor-
to varying degrees. On the other hand, rect responding, but only when they were
the subjects who did show equivalence asked in posttrial probes rather than in
used relational phrases most of the time. postsession interviews. Moreover, the
In a second experiment, Wulfert et al. button-press behavior of the I l-year-olds
(1991) controlled their subjects' verbal corresponded to their accurate descrip-
histories concerning the experiment by tions. And, although the button-press be-
pretraining responding either to the com- havior of the 4.5- and 7-year-olds was
pound stimuli or to the relation between fairly accurate, their descriptions were
stimuli. The results were as expected: more primitive than those ofthe I 1 -year-
Subjects trained to respond to stimulus olds. For example, although the 11-year-
compounds did not show equivalence olds gave contingency-specifying state-
during test trials, whereas subjects trained ments, such as, "You must count up to
to respond to the relation did show n (to get very good)," the 4.5-year-olds
equivalence. Moreover, subjects in both emitted phrases like, "press on the but-
groups continued to respond verbally ton," which specifies only behavior. In
during training and testing as they were contrast, the 7-year-olds said things like,
taught to do during pretraining. "You must press very hard," another CSS
Regarding the present discussion of which, as the authors pointed out, re-
verbal FASs, the Wulfert et al. (1 99 1) sulted incidentally in more successful re-
study is important mostly because it sponding; that is, the harder the button
demonstrates that the presence of certain press, the more likely it will be of longer
verbal stimuli, namely, those that specify duration. These results are interesting not
relations between stimuli (i.e., CSSs), is only because they reveal something about
DISCRIMINATIVE AND FUNCTION-ALTERING EFFECTS 19
the possible developmental course of Mistr (I 992) study is the only condition
FASs in children but also because they that suggests that the CSS may have been
reveal that even in humans who are prob- function altering and not evocative. In
ably capable of generating FASs, behav- this condition the toys, not the CSS,
ior can still be sensitive to the prevailing evoked the behavior of picking up the
contingencies (Baxter & Schlinger, 1990). toys. As predicted by a function-altering
Not surprisingly, the age of the child interpretation, "the child picked up the
is related to whether verbal stimuli will toys upon spotting the toys specified in
produce function-altering effects. Con- the statement" (Mistr, 1992). That is, the
sider the aforementioned study with 4- evocative function of the sight of the toys
and 5-year-old children by Mistr (1992). had been altered by the CSS. However,
This is perhaps the only experiment that the fact that the CSS reliably controlled
has attempted to separate evocative from the behavior of only 60% of the children
function-altering effects ofverbal stimuli. in this condition suggests that, when the
Mistr compared two sets of conditions, opportunity for immediate responding is
one in which the opportunity to respond prevented, CSSs such as the ones used
after a CSS was immediate and the other by Mistr may not perfectly control the
in which the opportunity to respond after behavior of 4- and 5-year-old children.
a CSS was delayed. Her instructions to This limitation may not be surprising
the children were similar to those used when one considers the complexity of the
in the Braam and Malott (1990) experi- CSSs used in this study. For example, in
ment. For example, in a condition that the condition with delayed deadlines but
specified no deadline, an immediate re- immediate reinforcers and a delayed op-
inforcer, and in which the opportunity to portunity to respond, the instruction to
respond was delayed, the instruction to the child was, "I'm going to put out some
the children was, "I'm going to put out toys for you to pick up later. I do not care
some toys for you to pick up later. I do if you pick them up or not. If you pick
not care if you pick them up or not. If up the toys before outside playtime, you
you pick up the toys, you can go to the can go to the Goodie Box when you're
Goodie Box when you're finished" (p. finished" (p. 20). This statement is fairly
22). Mistr found that, regardless of the complex. That it does not control the be-
specified deadlines or delays of reinforc- havior of 4- and 5-year-old children as a
ers, a majority of the children in the im- function-altering event is not surprising.
mediate-opportunity-to-respond condi- Taken as a whole, then, the experi-
tion responded within seconds of the ments described above on correspon-
stated CSS. This suggests that the CSSs dence training (e.g., Baer et al., 1988) and
evoked the relevant behavior. Interest- the studies by Braam and Malott (1990),
ingly, Mistr found that, with only one Mistr (1992), Pouthas et al. (1990), and
exception, when the opportunity to re- Wulfert et al. (1991) suggest that by the
spond was delayed, "Statements which age of I I years, humans are able to con-
specified a deadline and a reinforcer, ei- strue their own verbal FASs, and that for
ther immediate or delayed, did not reli- children as young as 4 or 5 years, simple,
ably control the behavior of 4 to 5 year but not complex, verbal stimuli are able
olds" (p. 49). The exception was the con- to alter stimulus functions reliably.
dition in which there was no deadline but
there was an immediate reinforcer. These THEORETICAL IMPLICATIONS:
results confirm the analysis of the Braam THE PROVENANCE OF
and Malott (1990) study offered above; FUNCTION-ALTERING CSSs
that is, their so-called "rules" were func-
tioning as evocative events-possibly The question that is most basic to the
EOs as they suggest-but not as function- present discussion is this: How do verbal
altering events. Incidentally, the no-de- stimuli acquire function-altering effects?
lay, immediate-reinforcer and delayed- When a verbal stimulus alters evocative
opportunity-to-respond condition in the discriminative-like functions it is easy to
20 HENRY D. SCHLINGER, JR.
appeal to a discrimination theory (e.g., novel manner. But what about verbal
Cerutti, 1989) to understand the prove- stimuli that are not CSSs; for example,
nance of the effects. Cerutti refers to the one-word stimuli that are function alter-
alteration of SD-like functions by verbal ing? What elements combine in these in-
stimuli as "instructed discrimination," stances? Our task as behavior analysts is
and he posits that the instructed stimuli to try to guess at the conditions necessary
are "6generalized discriminative classes" for the function-altering effect to happen.
whose "control over responses . . . gen- Let us briefly consider two types of
eralizes to new combinations of stimuli CSSs in an attempt to understand the dif-
in complex instructions" (p. 262). He ferences between a discriminative and a
gives the example of "Step on the brake function-altering approach to the prob-
pedal and come to a stop at a red light." lem. The first CSS is one that alters the
In Cerutti's terms, the control by red lights evocative (CS-like) effects of a stimulus,
over stopping is "instructed." In terms and the second is one that alters the evoc-
of the present analysis, the statement en- ative (SD-like) effects of a stimulus. In the
dows the red light with evocative control first example, an experimenter tells a sub-
over the behavior of stepping on the brake ject, "When you hear a bell you will feel
pedal. Although these descriptions sound a shock." Several minutes later a bell rings
like they are different ways of saying the and the subject's heart rate immediately
same thing, the real difference goes to the increases. This effect resembles, indeed,
main point of the present paper. Accord- mimics, respondent conditioning, but the
ing to Cerutti's discrimination theory of effect was achieved entirely as a result of
rule-governed behavior, each of the con- the CSS. Alessi (1992) has referred to this
stituent stimulus elements in the state- effect as (verbal) analogue respondent
ment controls one property of the re- conditioning. In essence, the bell and the
sponse, for example, the response itself, shock have been made equivalent by be-
its location, and its stimulus occasion. ing "paired" or "associated" within the
The entire statement exerts its control by statement, formally a "contingency-spec-
combining previously but separately ifying stimulus." What kind of history is
conditioned parts of the whole. necessary for such an effect to occur? The
Although the present paper attempts question, of course, is ultimately an em-
to make the case for separating discrim- pirical one, but we may speculate about
inative from function-altering effects of the answer in the absence of direct em-
verbal stimuli, an interpretation of rules pirical evidence. First, it seems that the
according to a discrimination theory re- subject must have independently heard
mains reasonable. After all, this sort of bells and felt some kind of electric shock
interpretation uses fundamental behav- before. Second, we might assume that the
ioral principles as a basis for the inter- subject can tact bells and shock; that is,
pretation of complex behavior (e.g., given the events themselves, the person
Palmer, 199 1; Schlinger, 1992). And, can say "bell," or "shock." Third, we
Cerutti's (1989) discrimination theory of might ask whether the person can iden-
rule-governed behavior is an admirable tify these events when the word "bell,"
attempt following from Skinner's inter- or "shock," is heard, for example, by
pretation of rules as SDS. However, as pointing toward the correct object. This
stated previously, there are other func- latter type of functional relation has been
tion-altering effects that do not seem eas- termed a stimulus-selection-based lan-
ily interpretable in operant discrimina- guage (Michael, 1985). We might also
tion terms. Perhaps there is more than want to know whether a matching-to-
one process at work, or perhaps there is sample repertoire is necessary or suffi-
a unitary process that is more than op- cient, especially because this type of rep-
erant discrimination. Certainly Cerutti ertoire forms the basis of the research on
must be correct in suggesting that verbal stimulus equivalence and stimulus
stimuli produce their effect in part be- equivalence has been suggested to be an
cause their elements combine in some important component in language func-
DISCRIMINATIVE AND FUNCTION-ALTERING EFFECTS 21
tion (Hayes, 1986). And, of course, we controls the behavior as an SD or an EO.
might want to know whether equivalence That is, it evokes the behavior (in com-
can already be demonstrated between ob- bination with the sight of the oven) be-
jects and the verbal responses (tacts) cause of a relevant (even generalized) his-
evoked by them (e.g., Sidman, 1971). tory. But what about the function of,
Whether verbal behavior is necessary for "When the timer goes off.. ."? In es-
stimulus equivalence (cf. Wulfert et al., sence, the timer and the command, "Take
1991) or stimulus equivalence is neces- the cake out of the oven," have been ren-
sary for "true" verbal behavior (Deva- dered functionally equivalent and, as a
ney, Hayes, & Nelson, 1986; Hayes, result, the evocative control by the state-
1986), it is obvious that verbal stimuli, ment is transferred to the sound of the
whether in the form of CSSs or not, can timer. Calling the statement an FAS sim-
alter stimulus function. Perhaps this ef- ply describes what happens, namely, that
fect is an emergent phenomenon. If so, the CSS alters the evocative-discrimi-
maybe it is what is truly unique about native-like -functions of the timer. What
verbal stimuli. is needed are empirical investigations into
In the second example, a mother tells how it happens.
her child: "When the timer goes off, please
take the cake out ofthe oven." Two hours
later, the timer goes off and the child im- CONCLUSION
mediately takes the cake out of the oven. Perhaps the main strength of behavior
As a result of the CSS, the timer evokes analysis has been the discovery of basic
the behavior of removing the cake from units of analysis. According to Zeiler
the oven and, therefore, mimics the ef- (1986),
fects of operant discrimination training.
This effect may be called analogue dis- The fundamental units (operants, respondents, dis-
criminative operants) are the smallest entities that
crimination training (e.g., Alessi, 1992). display the full characteristics of adaptive behav-
But how does it happen? It is possible, ior.... Research can involve the variables deter-
but unlikely, that the child simply re- mining how generic classes are constructed and the
peated the request during the entire two- factors responsible for particular forms of coordi-
hour period, although the child probably nated behavior. (pp. 4-5)
did repeat it a couple of times. It is also One implication of the present paper is
possible that when the timer timed out, that such "generic classes" are construct-
the child immediately said, "Oh I must ed through processes that can be de-
take the cake out of the oven." In this scribed as function altering. In other
case the self-statement might function in words, all of the processes that produce
part like an SD. Even if this were true, we the long-term behavior change that lead
would still have to explain how the timer us to speak of learning can be called func-
acquired evocative control over the self- tion altering. A function-altering classi-
statement. Of course, although the pres- fication scheme may permit all condi-
ent paper has argued against this practice, tioning processes, the verbal events that
we could simply classify this function as mimic them, and other seemingly unre-
another instance of stimulus control and lated learning phenomena (e.g., imprint-
call the statement an SD for taking the ing) to be considered in a more unified
cake out of the oven. But if we resist this manner, and may well suggest a common
discrimination interpretation, how can underlying mechanism of behavior
we account for the function-altering ef- change. Not only does such a scheme have
fects of this CSS? One approach is to an- important implications for how behavior
alyze the formal properties of the CSS analysts talk about their subject matter,
under the assumption that an under- but it also enables them to answer more
standing of each part might reveal some- effectively charges by some cognitively
thing about control by the whole. For oriented psychologists (e.g., Brewer, 1974;
example, the command, "Take the cake Chomsky, 1959) that behavior analysis
out of the oven," presumably already is unable to account for complex behav-
22 HENRY D. SCHLINGER, JR.
ioral processes, especially those involv- Brewer, W. F. (1974). There is no convincing ev-
ing language. idence for operant or classical conditioning in
In conclusion, the fact that verbal adult humans. In W. B. Weimer & D. S. Palerno
(Eds.), Cognition and the symbolic processes (pp.
stimuli can have discriminative (or re- 1-42). Hillsdale, NJ: Erlbaum.
inforcing) functions is indisputable. But Brownstein, A. J., & Shull, R. L. (1985). A rule
there are some events, including many for the use of the term, "Rule-governed behav-
that are verbal, that can mimic the effects ior." The Behavior Analyst, 8, 265-267.
of both respondent and operant condi- Buskist, W. F., Bennett, R. H., & Miller, H. L.
(1981). Effects of instructional constraints on
tioning by altering the functions of other human fixed-interval performance. Journal ofthe
events. Many behavior analysts agree that Experimental Analysis ofBehavior, 35, 217-225.
rule-governed behavior, however they Catania, A. C., Matthews, B. A., & Shimoff, E.
define it, seems to endow the human spe- (1990). Properties of rule-governed behavior and
their implications. In D. E. Blackman & H. Le-
cies with some very special capacities in- jeune (Eds.), Behavior analysis in theory and
deed. However, ifbehavior analysts con- practice: Contributions and controversies (pp. 215-
tinue to be interested only in the possible 230). Hove, England: Erlbaum.
discriminative (or motivational) effects Cerutti, D. T. (1989). Discrimination theory of
of verbal stimuli, they might miss what rule-governed behavior. Journal of the Experi-
mental Analysis ofBehavior, 51, 259-276.
may be most unique about such stimuli. Chomsky, N. (1959). Review of Verbal Behavior
Some behavior analysts have recognized by B. F. Skinner. Language, 35, 26-58.
that a more complex analysis may be Danforth, J. S., Chase, P. N., Dolan, M., & Joyce,
called for (e.g., Hayes, 1986; Hayes, Koh- J. H. (1990). The establishment of stimulus
control by instructions and by differential rein-
lenberg, & Hayes, 1991). The present forcement. Journal of the Experimental Analysis
paper concurs and suggests that the of Behavior, 27, 87-1 12.
complex analysis must include the func- Deacon, J. R., & Konarski, E. A. (1987). Corre-
spondence training: An example ofrule-governed
tion-altering nature of stimuli and the behavior?
processes that underlie their function-al- Journal of Applied Behavior Analysis,
20, 391-400.
tering effects. Devaney, J. M., Hayes, S. C., & Nelson, R. 0.
(1986). Equivalence class formation in language-
REFERENCES able and language-disabled children. Journal of
the Experimental Analysis of Behavior, 46, 243-
Alessi, G. (1992). Models of proximate and ul- 257.
timate causation in psychology. The American Hayes, S. C. (1986). The case of the silent dog-
Psychologist, 47, 1359-1370. Verbal reports and the analysis of rules. A review
Ayllon, T., & Azrin, N. H. (1964). Reinforcement of Ericson and Simon's ProtocolAnalysis: Verbal
and instructions with mental patients. Journal of Reports as Data. Journal of the Experimental
the Experimental Analysis of Behavior, 7, 327- Analysis of Behavior, 45, 351-363.
331. Hayes, S. C., Kohlenberg, B. S., & Hayes, L. J.
Baer, R. A., Detrich, R., & Weninger, J. M. (1988). (1991). The transfer of specific and general con-
On the functional role of the verbalization in cor- sequential functions through simple and condi-
respondence training procedures. Journal of Ap- tional equivalence relations. Journal of the Ex-
plied Behavior Analysis, 21, 345-356. perimental Analysis of Behavior, 56, 119-137.
Baron, A., & Galizio, M. (1983). Instructional Malott, R. W. (1989). The achievement ofevasive
control of human operant behavior. The Psycho- goals: Control by rules describing contingencies
logical Record, 33, 495-520. that are not direct acting. In S. C. Hayes (Ed.),
Baxter, G. A., & Schlinger, H. (1990). Perfor- Rule-governed behavior: Cognition, contingen-
mance ofchildren under a multiple random-ratio cies, and instructionalcontrol(pp. 269-322). New
random-interval schedule of reinforcement. York: Plenum.
Journal ofthe ExperimentalAnalysis ofBehavior, Matthews, B. A., Shimoff, E., Catania, A. C., &
54, 263-271. Sagvolden, T. (1977). Uninstructed human re-
Blakely, E., & Schlinger, H. (1987). Rules: Func-
tion-altering contingency-specifying stimuli. The sponding: Sensitivity to ratio and interval con-
BehaviorAnalyst, 10, 183-187. tingencies. Journal of the Experimental Analysis
Boren, J. J., & Devine, D. D. (1968). The repeated of Behavior, 27, 453-467.
acquisition of behavioral chains. Journal of the Michael, J. (1980). The discriminative stimulus
ExperimentalAnalysis ofBehavior, 11, 651-660. or SD. The Behavior Analyst, 3, 47-49.
Braam, C., & Malott, R. W. (1990). "I'll do it Michael, J. (1982). Distinguishing between dis-
when the snow melts": The effects of deadlines criminative and motivational functions of stim-
and delayed outcomes on rule-governed behavior uli. Journal of the Experimental Analysis of Be-
in preschool children. The Analysis of V'erbal Be- havior, 37, 149-155.
havior, 8, 67-76. Michael, J. (1983). Evocative and repertoire-al-
DISCRIMINATIVE AND FUNCTION-ALTERING EFFECTS 23
tering effects of an environmental event. The Schlinger, H., & Blakely, E. (1987). Function-al-
Analysis of Verbal Behavior, 2, 19-21. tering effects of contingency-specifying stimuli.
Michael, J. (1985). Two kinds of verbal behavior The Behavior Analyst, 10, 41-45.
plus a possible third. The Analysis of Verbal Be- Schlinger, H. D., Blakely, E., Fillhard, J., & Poling,
havior, 3, 1-4. A. (1991). Defining terms in behavior analysis:
Michael, J. (1986). Repertoire-altering effects of Reinforcer and discriminative stimulus. The
remote contingencies. The Analysis of Verbal Be- Analysis of Verbal Behavior, 9, 153-161.
havior, 4, 10-18. Sidman, M. (1971). Reading and auditory-visual
Mistr, K. (1992). The evocative and repertoire- equivalences. Journal ofSpeech and Hearing Re-
altering effects of contingency-specifying stimuli. search, 14, 5-13.
Unpublished master's thesis, University of North Skinner, B. F. (1957). Verbalbehavior. Englewood
Texas, Denton. Cliffs, NJ: Prentice-Hall.
Palmer, D. C. (1991). A behavioral interpretation Skinner, B. F. (1966). An operant analysis of prob-
of memory. In L. J. Hayes & P. N. Chase (Eds.), lem-solving. In B. Kleinmuntz (Ed.), Problem
Dialogues on verbal behavior (pp. 261-279). Reno, solving: Research, method, and theory (pp. 225-
NV: Context Press. 257). New York: Wiley.
Palmer, D. C., & Donahoe, J. W. (1992). Essen- Skinner, B. F. (1969). Contingencies of reinforce-
tialism and selectionism in cognitive science and ment: A theoretical analysis. New York: Apple-
behavior analysis. American Psychologist, 47, ton-Century-Crofts.
1344-1358. Vaughan, M. E. (1985). Repeated acquisition in
Pouthas, V., Droit, S., Jacquet, A.-Y., & Wearden, the analysis of rule-governed behavior. Journal
J. H. (1990). Temporal differentiation of re- ofthe Experimental Analysis ofBehavior, 27, 175-
sponse duration in children of different ages: De- 184.
velopmental changes in verbal and nonverbal be- Wulfert, E., Dougher, M. J., & Greenway, D. E.
havior. Journal of the Experimental Analysis of (1991). Protocol analysis of the correspondence
Behavior, 53, 21-31. of verbal behavior and equivalence class for-
Schlinger, H. D. (1990). A reply to behavior an- mation. Journal of the Experimental Analysis of
alysts writing about rules and rule-governed be- Behavior, 56, 489-504.
havior. The Analysis of Verbal Behavior, 8, 77- Zeiler, M. D. (1986). Behavioral units: A histor-
82. ical introduction. In T. Thompson & M. D. Zeiler
Schlinger, H. D. (1992). Theory in behavior anal- (Eds.), Analysis and integration ofbehavioral units
ysis: An application to child development. Amer- (pp. 1-12). Hillsdale, NJ: Erlbaum.
ican Psychologist, 47, 1396-1410.