The Behavior Analyst 1989, 12, 35-44 No.

1 (Spring)

Logic, Reasoning, and Verbal Behavior

Dudley J. Terrell
Anacapa Sciences, Inc.
J. M. Johnston
Auburn University
This paper analyzes the traditional concepts of logic and reasoning from the perspective of radical
behaviorism and in the terms ofSkinner's treatment of verbal behavior. The topics covered in this analysis
include the proposition, premises and conclusions, logicality and rules, and deductive and inductive
reasoning.
Key words: verbal behavior, reasoning, logic, induction, deduction, proposition

Skinner outlined a conceptual frame- dated by Creel (1987) and Schnaitter

work for the experimental analysis of (1980). Human logicality and reasoning
verbal behavior in 1957 with the publi- have been topics of investigation in psy-
cation of his book Verbal Behavior. Ex- chology for decades (for reviews, see Ev-
perimental research based explicitly on ans, 1982; Falmagne, 1975; Revlin &
this framework has emerged only in re- Mayer, 1978; Wason & Johnson-Laird,
cent years, however. This sparse body of 1968). To the degree that logic and rea-
work has included research on the verbal soning involve verbal behavior, both in-
operant classes of mands and tacts (Hall terpretative and experimental analyses
& Chase, 1986; Lamarre & Holland, based on Skinner's framework are there-
1985; Simac & Bucher, 1980), echoics fore appropriate. In the present paper,
(Boe & Winokur, 1978a, 1978b; Neville, the concepts of logic and reasoning are
1968), and intraverbals (Braam & Poling, analyzed in this conceptual context in or-
1983; Chase, Johnson, & Sulzer-Azaroff, der to facilitate experimental treatments
1985; Poon & Butler, 1972). In addition, of these topics. First, the concept of prop-
research on the independence of speaking osition is described and interpreted in be-
and listening (Lee, 1981) and on self-ed- havioral terms. Then, premises and con-
iting (Hyten & Chase, 1986) has been clusions, rules of logic, deductive
based on Skinner's work. This concep- reasoning, and inductive reasoning are
tual framework has also been utilized in analyzed as verbal behavior.
analyses of verbal discourse in group psy-
chotherapy (McLeish & Martin, 1975), THE PROPOSITION
maladaptive verbal behavior of the psy- The proposition is a basic component
chotherapy client (Glenn, 1983), audi- of logic and reasoning. The proposition
tory hallucinations (Bums, Heiby, &
Tharp, 1983), and instructional design has traditionally been conceptualized in
(Johnson & Chase, 1981). the following manner. There are acts in
Skinner devoted a chapter of his book which one may engage that require some
to an analysis of logical and scientific ver- sort of object for their execution. These
bal behavior (Skinner, 1957, chap. 18), acts are depicted by the transitive verbs.
and this general analysis has been eluci- For example, in order to hit, there must
be something to hit. This something, the
object of the action, is depicted by the
The authors would like to thank Bill Buskist, accusative of the transitive verb. Some
Peter Harzem, and Jorge de Oliveira-Castro for their
helpful comments on an earlier draft of this paper. transitive verbs depict what have cus-
Requests for reprints should be sent to Dudley J. tomarily been called acts of thinking (e.g.,
Terrell, P.O. Box 489, Fort Rucker, AL 36362. to know, believe, or assume). To know,
35
36 DUDLEY J. TERRELL & J. M. JOHNSTON
believe, or assume requires knowing, be- cussion of predication (1957, pp. 334-
lieving, or assuming something. These 335).
somethings, the accusatives of the acts of The concept of tact relates verbal be-
thinking, have come to be called prop- havior to nonverbal characteristics ofthe
ositions by logicians. In other words, a environment (e.g., objects and properties
proposition is what it is that one can of objects). A substantial portion of the
know, believe, think, judge, assume, human environment, however, includes
opine, and so on (Ryle, 1971). When peo- verbal stimuli (e.g., the verbal behavior
ple state or express what it is that they of other individuals). Verbal behavior re-
know, think, or assume, they have emit- lated to these aspects of the environment
ted behavior that can be analyzed in is called intraverbal (see Skinner, 1957,
Skinner's framework (1957). The re- pp. 71-78). Sometimes, subjects and
mainder of this section reinterprets the predicates of propositions may be related
traditional concept of the proposition to the environment as intraverbals, rath-
from this viewpoint. er than as tacts. The emission of such a
To begin this analysis, it is important response tacts the relation between the
to consider the behavior-environment intraverbals and their stimulating cir-
relation that Skinner called the tact ( 1 9 5 7, cumstance (e.g., "'John' is a proper
chap. 5) because it will later be shown noun").
that tacts and propositions are closely re- Finally, it is not unusual for the prop-
lated. A particular form of verbal re- osition to include a quantifying autoclitic
sponse is consistently reinforced in the (see Skinner, 1957, pp. 329-330). Under
presence of a particular object or event certain circumstances, the tact "swan"'
(e.g., a ball) or a property of an object or becomes "the swan" or "a swan." Other
event (e.g., its roundness). If the object circumstances may evoke the responses
or event controls the response, it is typ- "some swans," "all swans,"' or "no
ically called a pure tact. If some property swans." Skinner asserted that, as auto-
of the object or event controls the re- clitic components, these responses do not
sponse, it is called an abstract tact. There- modify the subject of the proposition. In-
fore, any given object may evoke several stead, their effect is to modify the reac-
verbal responses, some that are pure tacts tion of the listener to the responses they
and others that are abstract tacts con- accompany. "All" is "more appropriate-
trolled by the object's properties. ly taken as equivalent to always or always
Rather than simply emitting the dif- it is possible to say" (Skinner, 1957, p.
ferent forms of tacts separately (e.g., 329). Schoenfeld (1969) has noted a sim-
"ball," "round"), an additional response ilarity between the universal proposition
is often emitted that connects the tacts (e.g., "All swans are white") and the be-
(e.g., "The ball is round"). Skinner called havioral tendency toward generalization.
this additional response, "is," the asser- Such a response cannot possibly be under
tive autoclitic and stated that its function the control of all the swans in the uni-
is to enjoin the listener to "accept a given verse. Yet, enough exposure to reinforc-
state ofaffairs" (1957, p. 326). One might ing consequences in the presence of vari-
say that the roundness of the ball is the ations of stimulus conditions may result
state of affairs asserted by the speaker. It in highly generalized stimulus control.
may be more useful to consider that the The universal proposition tacts the re-
proposition, "The ball is round," tacts a lation between stimuli, responses, and
relation between the pure and abstract generalization of stimulus control.
tacts "ball" and "round.'" It is not the In summary, the proposition is a com-
roundness of the ball that evokes the en- plex verbal response that comprises tacts
tire response, "The ball is round." It is or intraverbals modified by particular au-
the relations between objects and each of toclitics, such as "is," "all," or "some."
their properties, as well as the respective As a unit of behavior, the proposition
tacts controlled by them, that evoke the functions as a tact in that it is controlled
utterance of propositions. This is essen- by the objects of simple tacts or intra-
tially what Skinner proposed in his dis- verbals and the relations between verbal
 LOGIC, REASONING, AND VERBAL BEHAVIOR 37
behavior and its environmental control. two ways. First, it the listener accepts or
In other words, when we say, "The ball acts upon the proposition as a verbal
is round," we tact not only the ball and stimulus, it has been effective. Second, if
its roundness, but our tendency to pred- the behavior of accepting it produces
icate roundness to the ball (Skinner, 1957, reinforcing consequences for the listener,
pp. 334-335). the proposition is again effective. A pre-
Having described the speaker's verbal cise definition of truth as an epistemo-
behavior involved in uttering a propo- logical issue is not required by a behav-
sition, we may now examine the effects ioristic analysis of the utterance of
of such behavior on a listener. Ifbehavior propositions. According to Zuriff(I 980):
produces reinforcing consequences, we Because of a specific phylogenetic and cultural his-
may say that it is effective. Responses tory, humans have evolved so that they are affected
producing consequences that reduce the in certain ways by verbal behavior, that is they
likelihood of future occurrences of a re- believe certain verbal behavior to be true. For the
sponse class may be considered ineffec- most part they do so without applying any explicit
tive. This argument is analogous to the criteria of truth. (p. 348).
adaptiveness of characteristics exhibited Exactly how and why any given propo-
during the evolution of a species (Skin- sition comes to be accepted as true by a
ner, 1984a). The verbal utterance of a listener is a problem for empirical psy-
proposition may or not produce rein- chology (Popper, 1959).
forcing consequences. The emission of a
proposition may be reinforced when a REASONING
listener accepts or believes it. According Premises and Conclusions
to Skinner, "Our belief in what someone
tells us is ... a function of, or identical If a listener's behavior of accepting a
with, our tendency to act upon the verbal proposition reinforces a speaker's behav-
stimuli which he provides" (1957, p. 160). ior, the speaker may respond in ways that
Therefore, the utterance of a proposition will increase the probability of that rein-
may be effective (i.e., accepted or be- forcing consequence. This supplemen-
lieved by a listener) regardless of its cor- tary behavior may be considered verbal
respondence to any state of affairs. Like- to the extent that it is also reinforced by
wise, an utterance may be ineffective (i.e., the listener's response. For example, a
rejected or ignored by a listener) regard- speaker may increase the probability of
less of its truth. a listener accepting a proposition (i.e.,
Additionally, the listener's behavior in reinforcing the speaker's behavior) by
response to the proposition may also be emitting other propositions that are more
classified as effective or ineffective. Again, readily accepted by the listener. In the
behavior that produces reinforcement is traditional language of logicians, these
considered effective. Whether the behav- more readily accepted propositions are
ior of accepting or rejecting a proposition called premises, and the proposition
produces reinforcing consequences may whose acceptability is subsequently en-
have something to do with the "truth" hanced is called the conclusion. The ac-
of the statement. Skinner stated that "a ceptance of the premises is assumed to
proposition is true to the extent that with be at a greater strength in the listener's
its help the listener responds effectively behavioral repertoire than the accep-
to the situation it describes" (1974, p. tance of the unpremised conclusions. In
235). This statement applies to tacts in some cases, the speaker may preface the
general. Again, according to Skinner, premises with the mand "Suppose that
"behavior in the form of the tact works ... ." This utterance may then facilitate
for the benefit of the listener by extending the acceptance of the conclusion.
his contact with the environment, and In colloquial language, the speaker is
such behavior is set up in the verbal com- attempting to prove a point, support a
munity for this reason" (1957, p. 85). conclusion, or convince the listener of a
In summary, the utterance of a prop- proposition's truth, viability, or possi-
osition has been defined as effective in bility. For example, a prosecutor may as-
38 DUDLEY J. TERRELL & J. M. JOHNSTON
sert or propose that a defendant is guilty the occasion for the emission of other
of some crime. The acceptance of this verbal responses (i.e., conclusions are
proposition by the members of a jury will drawn and theories are proposed). Sim-
be reinforcing to the prosecutor. By itself, ilarly, some scientists may respond to a
the single proposition may have a min- number of different theories by emitting
imal effect on the jury members' behav- a proposition that must be true if the the-
ior. Therefore, the prosecutor produces ories (or premises) are true (i.e., a hy-
other assertions that members of the jury pothesis). The research scientist then goes
readily accept as true. These assertions on to test this proposition experimental-
essentially describe the "evidence" for the ly.
conclusion (e.g., defendant was present Nonscientific verbal behavior may also
at scene of crime; weapon was found in involve inferring or drawing conclusions.
defendant's possession; defendant's alibi For example, a salesperson may provide
is weak, etc.). If these premises have the reasons for buying a particular product
intended effect on the jury, it is more at a particular time. The reasons may be
likely that a verdict of guilty will be re- acceptable to a potential buyer and ul-
turned. Thus, the members of the jury timately occasion the emission of the
will have accepted the prosecutor's orig- conclusion, "Now is the time to buy this
inal proposition and probably reinforced item."
that verbal behavior. In summary, a speaker may emit state-
The emission of propositions in such ments in ways that influence a listener's
a manner constitutes a pattern of verbal acceptance or emission of other state-
behavior that is called reasoning. Speak- ments. Specifically, the emission of the
ers reason with listeners when they emit premises affects the acceptance or emis-
verbal behavior that alters the probabil- sion of conclusions. Two senses of such
ity of the listener accepting certain prop- verbal reasoning have been described in
ositions. In the previous example, the terms of the behavior of speaker and lis-
prosecutor provides reasons for the jury tener. In one sense, a listener's behavior
to accept the proposition that the de- of accepting a conclusion is modified
fendant is guilty. Speakers are also said when a speaker emits more readily ac-
to reason when they emit propositions cepted premises. In another sense, a lis-
with higher levels of acceptability (prem- tener's behavior of emitting a conclusion
ises) and thereby produce verbal stimuli is modified when premises are emitted
that facilitate the emission of a conclud- in a particular manner. In either case, the
ing proposition. Premises may be emit- speaker and listener may be two (or more)
ted in spoken or written form and may different people or may be the same per-
occur in different orders or sequences. son. One may "prove" a point to oneself
The products of this behavior, the verbal as well as to someone else. Likewise, one
stimuli, may then set the occasion for may draw one's own inferences, as well
new propositions to be emitted. We say as inspire someone else to draw them.
that conclusions have been drawn or in- The manner in which premises have
ferred from the premises. Acceptance of such effects on conclusions is the topic of
conclusions or, rather, effective behavior the remaining sections of this paper. Af-
in response to conclusions may reinforce ter a discussion of rules as descriptions
the pattern of behavior that produced of regularity in patterns of behavior, we
them. will present analyses of the concepts of
Examples of this kind of behavior are deductive and inductive reasoning.
ubiquitous in science. The behavior of
the scientific theoretician includes a "set Logicality and Rules
of manipulative responses directed, not
at the natural subject matter of the sci- The temporal sequence in which events
ence, but at the verbal record of that sub- occur may be described as a pattern of
ject matter, the data" (Schnaitter, 1980, events. As a subset of physical events,
p. 159). Verbal responses tacting the re- behavior occurs in temporal sequences
sults of experimental manipulations set or patterns. Some patterns come to be
 LOGIC, REASONING, AND VERBAL BEHAVIOR 39
recognized as regularities or consisten- however, additional control exerted by
cies in nature. Verbal behavior may tact prevailing social contingencies is not to
patterns of events in the same way that be understated.
it tacts single events. Verbal behavior that Furthermore, the degree of control ex-
tacts consistent patterns in nature gen- erted by a verbal statement of contingen-
erally results in statements that are typ- cies may shift in the development and
ically called rules. For example, certain refinement of behavior (Buskist & Miller,
forms of verbal behavior described as 1986; although cf. Hayes, Brownstein,
grammatical are patterns of behavioral Haas, & Greenway, 1986). An individual
regularities with which we are familiar as learning a second language may first learn
rules. As verbal responses, rules tact not the rules of grammar for that language,
specific events, but the relationships and for some time those rules may be
among the events (i.e., the consistent or meticulously consulted before each ut-
regular patterns in which the events oc- terance. Once fluent, however, the speak-
cur). er usually no longer consults the rules
These verbal responses (rules) may before speaking. Rather, the verbal be-
originate as descriptions of regularity; havior comes under the control of the
however, they may very often become foreign verbal community's reinforce-
prescriptive by aiding in the verbal con- ment contingencies.
trol of human behavior. For example, the With regard to reasoning, certain con-
rules of grammar describe some consist- sistencies may exist in the way premises
encies in the reinforcement practices of might be emitted in altering the proba-
members of a verbal community (Skin- bility that a conclusion will be emitted
ner, 1957). These descriptions of rein- or accepted. Certain patterns or arrange-
forcement contingencies do not neces- ments of premises are more effective than
sarily affect the events they describe. They others in facilitating the emission or en-
are verbal descriptions, not the contin- hancing the acceptability of conclusions.
gencies themselves. However, the rules A description of these consistencies may
may be "helpful in instruction and in be stated in terms of the reinforcement
maintaining verbal behavior in conform- practices of a particular verbal commu-
ity with the usages of the community" nity, such as the logical/scientific or the
(Skinner, 1984b, p. 585). lay verbal community. Furthermore, de-
It is important to distinguish between scriptions of consistencies in reinforce-
rule-governed or rule-following behavior ment practices with regard to the emis-
and rule-characterized behavior. Al- sion and acceptance of propositions might
though any given set of responses may be called the rules of logicality.
be described as corresponding to some The rules oflogic, sometimes called the
rule, the rule, as a description of the con- rules of deductive inference (which are
tingency, does not necessarily control the discussed more extensively. in the next
behavior. The contingency itself may section), may be considered a more for-
control the behavior. For example, an malized subset of the rules of logicality.
individual may be described as speaking At this point, the rules of logicality may
grammatically, although the verbal be- be described as the broad set of verbal
havior is entirely under the control of the responses that tact regularities in patterns
prevailing social contingencies. This be- of proposition utterances (see examples
havior may be called rule-corresponding below). When the emission of proposi-
or rule-characterized, but it is not rule- tions corresponds to such rules, the prob-
following or rule-governed. On the other ability of reinforcement (i.e., the accep-
hand, the cautious writer or a speaker of tance of the conclusion or the emission
a foreign language may consult a rule of an effective conclusion) is maximal.
book, such as a style manual or diction- Since arranging and emitting proposi-
ary, before emitting a statement. Such tions is a verbal process, the rules of log-
verbal behavior may be considered rule- icality may, along with the rules of gram-
following to the extent that the verbal mar, be considered a subset of the rules
stimuli, the rules, control the behavior; of language (i.e., the descriptions of con-
40 DUDLEY J. TERRELL & J. M. JOHNSTON

sistencies and regularities in the general and implication. A conjunction is a com-

reinforcement practices of a verbal com- bination of two propositions. The verbal
munity). response "and" functions to connect two
Again, any given set of responses, such responses that may just as easily have
as those constituting a logical discourse, been emitted separately (e.g., "It is Fri-
may be considered rule-characterized or day and I am writing"). The emission of
rule-governed depending on the nature two propositions connected in such a
of the controlling variables. A student of manner constitutes a more complex
logic may consult the rules of deductive propositional response, if only for the ef-
inference to draw a conclusion from a set fect it has on the listener. If a listener
of premises; yet, an experienced logician accepts any two simple propositions, it
may derive conclusions from premises is most probable that a conjunctive prop-
much in the same way that the experi- osition (i.e., the two simple propositions
enced poet thinks in a particular poetic connected by the response "and") will be
meter (see Skinner, 1957, p. 422). In fact, accepted. This is the deductive inference
it seems likely that logical verbal behav- rule called conjunction. Alternatively, if
ior emerges or develops in the normal a listener accepts a conjunctive propo-
course of human development under sition, the behavior of emitting any one
processes similar to those in the devel- of the conjuncts will most likely be ac-
opment of verbal behavior in general (e.g., cepted. This is the rule of simplification.
Inhelder & Piaget, 1958, 1969). Two propositions are said to be dis-
juncted when they are connected by the
response "or" (e.g., "Either class has been
Deductive Reasoning cancelled, or I'm in the wrong room"). If
a disjunctive proposition is accepted as
In the language of logicians, proposi- a premise, and if a second premise ne-
tions may be arranged in such a manner gating one of the disjuncts is also ac-
that the conclusion is necessarily true cepted, it is most probable that a conclu-
given premises that are true. That is, giv- sion in the form of the other disjunct will
en an arrangement of propositions that be accepted. This is the rule of disjunc-
may be said to correspond to a particular tive syllogism. Acceptance of the prem-
state of affairs, a concluding proposition ises "Class has been cancelled or I'm in
can be formulated that also corresponds the wrong room" and "I'm not in the
to the particular state ofaffairs. The form wrong room" facilitates the acceptance
and arrangements these propositions or emission of the conclusion "Class has
must take are characterized by the formal been cancelled."
rules of logic, sometimes called the rules Another pattern of behavior is de-
of deductive inference. scribed in the deductive rule called ad-
Behavioristically restated, patterns of dition. According to the rule of addition,
verbal response forms may consistently any single true proposition may be dis-
produce effective behavior on the part of junctively connected to any other prop-
the listener (i.e., acceptance or emission osition. This pattern of behavior may not
of effective conclusions). The rules of de- occur very often in ordinary discourse.
ductive inference describe the patterns of According to Braine, "If p is already es-
behavior that have a high probability of tablished, there is no reason to want to
reinforcement. As a subset of the rules of infer the weaker statement, p or q, which
logicality, these rules are verbal re- suggests doubt about p" (1 978, p. 14). It
sponses that tact the relations among be- may be true that anyone who accepts the
havioral events of emitting premises and simple proposition will also accept the
emitting and accepting conclusions (i.e., disjunctive proposition (although it is an
verbal reasoning). empirical issue), but it is unclear what
To illustrate, consider three ways that function such a pattern of verbal behav-
logicians describe the relations among ior might have.
propositions: conjunction, disjunction, An implication is another case of two
 LOGIC, REASONING, AND VERBAL BEHAVIOR 41

simple propositions connected by an ad- tive simply if the behavior can be char-
ditional verbal response. The responses acterized or described by these rules.
that make this connection take various
forms, such as "If. . . , then.. . " or "im-
plies." This kind of proposition may Inductive Reasoning
function as a tact ofintraverbal relations,
such as class inclusion (e.g., "If you have We have suggested that rules of de-
a cat, then you have a pet") or definition ductive inference are descriptions of re-
(e.g., "If one of the angles is 90 degrees, inforcement contingencies for effective
then it is a right triangle"). It may also sequences of proposition utterances.
tact a contingency (e.g., "If you are late, Premises can be constructed and ar-
then I'll leave without you") or a causal ranged in ways that may enhance the ac-
relation (e.g., "Combustion implies the ceptability of a conclusion, but that do
presence of oxygen"). If a listener accepts not correspond to the rules of deductive
such a proposition, and if the listener fur- inference. This kind of verbal behavior
ther accepts a simple proposition which may be called inductive reasoning. We
constitutes the antecedent of the impli- do not argue for a functional distinction
cation, it is most probable that the lis- between inductive and deductive pat-
tener will accept the simple proposition terns of verbal behavior. The distinction
which constitutes the consequent of the originated in the writings of logicians, but
implication. This is the rule of modus from a behavioral perspective the dis-
ponens. Acceptance of the proposition tinction seems formal or nominal in na-
that combustion implies oxygen and that ture. Behavior is called deductive rea-
combustion is present facilitates the ac- soning simply if it can be related to the
ceptance of the proposition that oxygen rules of deductive inference, whether le-
is present. Alternatively, if a listener ac- gitimate or fallacious. Although logicians
cepts an implication and further accepts find the definition ofinductive reasoning
a proposition negating the consequent, it difficult, behavior is generally labeled as
is most probable that the listener will ac- inductive if it leads to conclusions that
cept a conclusion negating the anteced- are only probably true. In other words,
ent. This is the rule of modus tollens. Ac- inductive reasoning is not defined in
ceptance ofthe premises that combustion terms of a set of formal rules. There are
implies oxygen and that oxygen is not consistencies in the patterns of behavior
present will facilitate the acceptance of called inductive reasoning, and these
the conclusion that combustion is not consistencies may be described as rules
present. from a behavioristic perspective.
Other rules ofdeductive inference (e.g., Consider, for example, two of J. S.
hypothetical syllogism, De Morgan's Mill's (1973/1843) methods of inductive
theorem, exportation) may be described inference. Mill maintained that consis-
in a similar manner. A person reasons tencies exist in the way one discovers and
deductively when the emission of prop- demonstrates causal relations in scientif-
ositions corresponds to these formal rules ic investigation. If one observes a com-
of logic. Again, deductive reasoning is mon variable in several otherwise dis-
not necessarily an instance of rule-gov- parate circumstances, that variable may
erned behavior. An individual's behav- be inferred to be the cause, or effect, of
ior may be partially under the control of the phenomenon under investigation.
verbal stimuli constituting these rules. This is Mill's Method of Agreement. If a
"Rules of evidence in a court of law re- number of people all exhibit some sim-
strict the verbal behavior of witnesses, ilar disease symptomatology, and these
the rules of chess restrict the movements people have no common history except
of the pieces, logical rules have a com- a deficit of fresh fruit and vegetables in
parable effect on the logician" (Skinner, their diet, one might infer that the lack
1957, p. 423). Evidence of rule-following of fresh fruit and vegetables is a cause of
is not required. The reasoning is deduc- their illness.
42 DUDLEY J. TERRELL & J. M. JOHNSTON
Mill's Method of Differences may be they may occasionally, if not frequently,
considered in a similar manner. If an produce reinforcement (i.e., effective be-
event occurs in the presence of some oth- havior ofaccepting the conclusion). These
er event, and never in its absence, one patterns of behavior are considered to be
may infer that the two events are causally cases of fallacious reasoning, although
related. This kind of causality may be sometimes what superficially appears to
more widely accepted as necessary con- be a case of illicit modus ponens (or tol-
ditionality. Combustion may occur in the lens) is actually deductively legitimate.
presence of oxygen (as well as additional For example, denying the antecedent or
necessary conditions), but never in the affirming the consequent of an implica-
absence of it. tion that functions as a definitional tact
Another inductive practice is reason- is deductively valid (e.g., "If 90 degrees,
ing by analogy. A number of events, ob- then right angle").
jects, or circumstances are described as Mill's methods, analogical reasoning,
having a number of common properties and some of the deductive fallacies are
or characteristics. If several of these examples of verbal behavior that may
events, objects, or circumstances have an have a high probability of reinforcement.
additional common characteristic, one The legitimacy of inductively derived
may infer by analogy that the remaining conclusions, in terms of correspondence
circumstances also have the additional with actual states of affairs, has been an
characteristic. For example, if John, Bob, epistemological controversy since the
and Paula are all graduates of the same time of Hume (1955/1748). The proba-
school and have all attained satisfying bility that such conclusions are accepted
careers, and Jane is also a graduate of that by a listener and the effects of various
school, one may conclude that Jane is propositional manipulations on that
likely to attain a satisfying career. This probability are empirical issues suitable
kind of reasoning may be the result of the for experimental psychology.
behavioral phenomenon of stimulus gen- If some pattern of behavior produces
eralization. If verbal responses are rein- more reinforcement than other patterns,
forced in the presence of a particular dis- we may expect that pattern to be of great-
criminative stimulus, other stimuli that er strength than other patterns in the be-
have physical characteristics or relations havioral repertoire of an individual. The
in common with the discriminative stim- behavior is explained or justified by re-
ulus may also control similar verbal re- ferring to the reinforcement process that
sponding. maintains it. To say that deductive rea-
Other patterns of verbal behavior in soning is justified by the rules of deduc-
which premises affect the acceptability of tive inference may mean nothing more
conclusions are described as logical fal- than that the behavior is adequately rein-
lacies. It is common for the affirmed con- forced in the verbal community, and the
sequent of an implication to increase the rules of inference describe the reinforce-
acceptability of an affirmed antecedent as ment contingencies. If patterns of induc-
a conclusion. Logicians call this pattern tive reasoning are similarly maintained
of discourse illicit modus ponens. For in- by the verbal community, they are sim-
stance, "If it rained last night, then the ilarly justified. We may call this justifi-
ground will be wet this morning. The cation psychological rather than logical.
ground is wet, therefore it must have In other words, the psychological jus-
rained last night." Illicit modus tollens, tification of behavior lies in an under-
known as denying the antecedent, is sim- standing of reinforcement contingencies.
ilar. "If it rained last night, then the The reinforcement contingencies for in-
ground will be wet this morning. It did ductive behavior have been studied in
not rain last night, therefore the ground the context of generalization. In fact, "in-
is not wet." ductive inference" and "induction" are
Although these patterns do not corre- expressions that have been used tradi-
spond to the rules ofdeductive inference, tionally to describe the same phenomena
 LOGIC, REASONING, AND VERBAL BEHAVIOR 43
that we now call generalization from ex- of reasoning that have high probabilities
perience or stimulus generalization. Con- of reinforcement may be described by a
sider Mill's observation, "The child, who set of contingencies that might be called
having burnt his fingers, avoids to thrust the rules of logicality. Behavior that cor-
them again into the fire, has reasoned or responds to such rules, but is not nec-
inferred, though he has never thought of essarily under the control of verbal de-
the general maxim, Fire burns" (1973/ scriptions of the contingencies, has a high
1843, p. 188). Skinner described it in be- probability of producing reinforcement.
havioristic language: "The spread of ef- One subset of the rules of logicality is
fect to other stimuli is called generaliza- the set of rules of deductive inference.
tion or induction" (195 3, p. 132). Finally, Logicians say that correspondence to
Sidman explicitly stated, "Induction is a these rules guarantees the truth of con-
behavioral process.... Whether or not clusions given the truth of premises. The
we make an inductive inference, and the rules may describe patterns of reasoning
degree of tenacity to which we cling to that have a high probability of being ac-
that inference, will depend upon our be- cepted by a listener. There are no formal
havioral history (experience)" (1960, p. rules of inductive inference, but consis-
59). tencies in patterns of inductive reasoning
The patterns of nonverbal behavior can be described and related to basic
that may be called inductive reasoning or principles of behavior. Consistencies in
inference are the same patterns that are the reinforcement practices for verbal in-
explained in the body of knowledge con- ductive reasoning remain a topic for be-
stituting the science ofbehavior. Skinner havior analysis.
(1957) has proposed that verbal behavior An experimental analysis of the verbal
be analyzed in the same manner as non- behavior involved in logic and reasoning
verbal behavior, and there is no reason may uncover functional relations that will
to account for verbal inductive reasoning enhance a person's effectiveness with re-
in any other way. The present analysis gard to the reinforcing environment.
extends this approach to the deductive Some of the relevant behavioral issues
patterns that have been the subject of tra- have already been raised. How does log-
ditional logical investigation. icality develop in the human repertoire?
How do propositions develop or emerge
SUMMARY from simple tacts and intraverbals? What
factors control the acceptance of the
An analysis of verbal reasoning and premises of a logical discourse? How well
logical verbal behavior begins with an do the rules of deductive inference de-
analysis of the proposition. Stating prop- scribe the behavioral effects of deductive
ositions involves emitting verbal re- reasoning? What are the variables that
sponses that are related as tacts to their affect the acceptance (or emission) of in-
environmental circumstances and to typ- ductive conclusions? These are only a few
ical relations between similar circum- of the questions that the science of be-
stances and verbal behavior in general. havior might address in the study of rea-
This kind of verbal behavior may be con- soning and logical verbal behavior.
sidered effective ifit produces reinforcing
consequences, such as acceptance by a REFERENCES
listener.
Some propositions may be rendered Boe, R., & Winokur, S. (1978a). A procedure for
studying echoic control in verbal behavior. Jour-
more acceptable to a listener when they nal of the Experimental Analysis ofBehavior, 33,
accompany other, more readily accept- 495-520.
able propositions called premises. Verbal Boe, R., & Winokur, S. (1978b). Echoic control
reasoning involves altering the probabil- in conversational speech. Journal ofGeneral Psy-
chology, 99, 299-304.
ity that conclusions will be accepted or Braam, S. J., & Polling, A. (1983). Development
emitted by emitting and manipulating of intraverbal behavior in mentally retarded in-
premises. Any consistencies in patterns dividuals through transfer of stimulus control
44 DUDLEY J. TERRELL & J. M. JOHNSTON
procedures: Classification of verbal responses. tional independence of mands and tacts. Journal
Applied Research in Mental Retardation, 4, 279- of the Experimental Analysis of Behavior, 43, 5-
302. 19.
Braine, M. D. S. (1978). On the relation between Lee, V. L. (1981). Prepositional phrases spoken
the natural logic of reasoning and standard logic. and heard. Journal of the Experimental Analysis
Psychological Review, 85, 1-21. of Behavior, 35, 227-242.
Burns, C. E. S., Heiby, E. M., &Tharp, R. G. (1983). McLeish, J., & Martin, J. (1975). Verbal behavior:
A verbal behavior analysis of auditory halluci- A review and experimental analysis. Journal of
nations. The Behavior Analyst, 6, 133-143. General Psychology, 93, 3-66.
Buskist, W. F., & Miller, H. L. (1986). Interac- Mill, J. S. (1973/1843). A system of logic. In F.
tions between rules and contingencies in the con- F. McRae (Ed.), Collected works of John Stuart
trol of human fixed-interval performance. The Mill (Vols. 7 & 8). Toronto: University of To-
Psychological Record, 36, 109-116. ronto Press.
Chase, P. N., Johnson, K. R., & Sulzer-Azaroff, B. Neville, M. H. (1968). Effects of oral and echoic
(1985). Verbal relations within instruction: Are responses in beginning reading. Journal of Edu-
there subclasses of the intraverbal? Journal ofthe cational Psychology, 59, 362-369.
ExperimentalAnalysis ofBehavior, 43, 301-313. Poon, W., & Butler, K. G. (1972). Evaluation of
Creel, R. (1987). Skinner on science. In S. Modgil intraverbal responses in five- to seven-year old
& C. Modgil (Eds.), B. F. Skinner: Consensus and children. Journal of Speech and Hearing Re-
controversy (pp. 103-111). New York: Falmer. search, 15, 303-307.
Evans, J. St. B. T. (1982). The psychology of de- Popper, K. R. (1959). The logic of scientific dis-
ductive reasoning. London: Routledge & Kegan covery. New York: Basic Books.
Paul. Revlin, R., & Mayer, R. E. (1978). Human rea-
Falmagne, R. J. (1975). Reasoning: Representa- soning. Washington, DC: V. H. Winston.
tion and process in children and adults. Hillsdale, Ryle, G. (1971). Collected papers (Vol. 2). Lon-
NJ: Lawrence Erlbaum. don: Hutchinson.
Glenn, S. S. (1983). Maladaptive functional re- Schnaitter, R. (1980). Science and verbal behav-
lations in client verbal behavior. The Behavior ior. Behaviorism, 8, 153-160.
Analyst, 6, 47-56. Schoenfeld, W. N. (1969). J. R. Kantor's Objective
Hall, G. A., & Chase, P. N. (1986, May). Func- psychology ofgrammar and Psychology and logic:
tional transfer between mands and tacts. Poster A retrospective appreciation. Journal ofthe Ex-
presented at the meeting of the Association for perimental Analysis of Behavior, 12, 329-347.
Behavior Analysis, Milwaukee, WI. Sidman, M. (1960). Tactics ofscientific research.
Hayes, S. C., Brownstein, A. J., Haas, J. R., & New York: Basic Books.
Greenway, D. E. (1986). Instructions, multiple Simac, J., & Bucher, B. (1980). Development of
schedules, and extinction: Distinguishing rule- spontaneous manding in language deficient chil-
governed from schedule-controlled behavior. dren. Journal ofApplied Behavior Analysis, 13,
Journal ofthe ExperimentalAnalysis ofBehavior, 523-528.
46, 137-147. Skinner, B. F. (1953). Science and human behav-
Hume, D. (1955/1748). An inquiryconcerninghu- ior. New York: The Free Press.
man understanding. New York: Liberal Arts Press. Skinner, B. F. (1957). Verbalbehavior. Englewood
Hyten, C., & Chase, P. N. (1986, May). Experi- Cliffs, NJ: Prentice-Hall.
mental analysis of self-editing. Poster presented Skinner, B. F. (1974). About behaviorism. New
at the meeting of the Association for Behavior York: Knopf.
Analysis, Milwaukee, WI. Skinner, B. F. (1984a). The phylogeny and on-
Inhelder, B., & Piaget, J. (1958). The growth of togeny of behavior. Behavioral and Brain Sci-
logical thinking from childhood to adolescence. ences, 7, 669-711.
New York: Basic Books. Skinner, B. F. (1984b). An operant analysis of
Inhelder, B., & Piaget, J. (1969). Theearlygrowth problem solving. Behavioral and Brain Sciences,
oflogic in the child. New York: W. W. Norton. 7, 583-613.
Johnson, K. R., & Chase, P. N. (1981). Behavior Wason, P. C., & Johnson-Laird, P. N. (1968).
analysis in instructional design: A functional ty- Thinking and reasoning. Middlesex, England:
pology of verbal tasks. The Behavior Analyst, 4, Penguin Books.
103-121. Zuriff, G. E. (1980). Radical behaviorist episte-
Lamarre, J., & Holland, J. G. (1985). The func- mology. Psychological Bulletin, 87, 337-350.