The Behavior Analyst 1998, 21, 1-12 No.

1 (Spring)

Making Sense of Sensitivity in the

Human Operant Literature
Gregory J. Madden
University of Vermont
Philip N. Chase and James H. Joyce
West Virginia University
Human operant behavior is often said to be controlled by different variables or govemed by different
processes than nonhuman operant behavior. Support for this claim within the operant literature comes
from data suggesting that human behavior is often insensitive to schedules of reinforcement to which
nonhuman behavior has been sensitive. The data that evoke the use of the terms sensitivity and
insensitivity, however, result from both between-species and within-subject comparisons. We argue
that because sensitivity is synonymous with experimental control, conclusions about sensitivity are
best demonstrated through within-subject comparisons. Further, we argue that even when sensitivity
is assessed using within-subject comparisons of performance on different schedules of reinforcement,
procedural differences between studies of different species may affect schedule performance in im-
portant ways. We extend this argument to age differences as well. We conclude that differences across
populations are an occasion for more precise experimental analyses and that it is premature to conclude
that human behavior is controlled by different processes than nonhuman behavior.
Key words: humans, sensitivity, schedules of reinforcement, concurrent schedules, interspecies
continuity

A great deal has been written about Lowe, 1996; Lowe, 1979). Frequently
differences between human and non- the new principles called for are those
human sensitivity to schedules of re- related to verbal behavior. For exam-
inforcement (e.g., Baron & Galizio, ple, Lowe and Home (1996) concluded
1983; Cerutti, 1989; Galizio, 1979; that "(a) the performance of verbally
Kaufman, Baron, & Kopp, 1966; able humans on schedules of reinforce-
Lowe, 1979; Shimoff, Catania, & Mat- ment, including concurrent schedules,
thews, 1981; Skinner, 1966). Some differed greatly from that observed in
have argued that because human and nonhuman species, and that (b) a key
nonhuman behavior is affected in dif- variable in bringing about these differ-
ferent ways by seemingly comparable ences was human subjects' ability to
schedule contingencies, different prin- specify the contingencies verbally and
ciples are required for accurate ac- to formulate their own rules for re-
counts of human and nonhuman behav- sponding" (p. 315). Consistent with
ior (e.g., Brewer, 1974; Home & this argument, a number of definitions
of rule-governed behavior have been
This work was supported by NIDA Grant 5- presented that include schedule insen-
ROI-DA06526-08. sitivity as a critical feature (Catania,
Portions of this article were written as the first Shimoff, & Matthews, 1989; Cerutti,
author's conceptual preliminary examination 1989; Shimoff et al., 1981).
while a doctoral student at West Virginia Uni- These arguments are important be-
versity. We are grateful to Margaret Vaughan cause they suggest that human behav-
and several anonymous reviewers who provided
insightful comments on the manuscript. James ior, at least that which is rule governed,
Joyce is now at the Continuous Learning Group, is not sensitive to changes in schedules
Jamesville, Virginia. of reinforcement. We find it difficult to
Address correspondence and reprint requests agree with this conclusion. Our review
to Gregory J. Madden, Human Behavioral Phar-
macology Laboratory, University of Vermont, of the human operant literature reveals
38 Fletcher Place, Burlington, Vermont 05401- two potentially conflicting definitions
1419. of the term sensitivity (we assume
1
2 GREGORY J. MADDEN et al.

throughout that sensitivity and insen- resembled the pause-respond patterns

sitivity describe opposite ends of a of nonhumans (a between-species com-
continuum of verbal responses that an parison). Similarly, B. Matthews, Shi-
experimenter makes in the presence of moff, Catania, and Sagvolden (1977)
a set of behavioral data). According to described human performances under
one definition, human behavior is multiple, concurrent, and Fl schedules
schedule sensitive only when it resem- as being insensitive when perfor-
bles schedule-sensitive nonhuman be- mances were divergent from response
havior. The other defines sensitivity as patterns typifying nonhuman behavior
a change in behavior following a con- under similar conditions. Like Lowe et
tingency change. We believe that when al., Matthews and colleagues suggested
these two definitions are pinpointed that observing-response procedures
and applied to the findings of the hu- were an effective means of improving
man operant literature, the conclusions human schedule sensitivity, "at least in
about sensitivity and insensitivity are the sense that they generate scalloping
not as clear as suggested by Lowe and in human Fl responding" (p. 454).
Home (1996) or others. Therefore, the When Matthews et al. discussed the re-
present paper will (a) distinguish more lation between insensitivity and in-
explicitly these two definitions of structed performances, they contended
schedule sensitivity, (b) review some that manipulating schedule contingen-
confusions apparently generated by cies can reveal the influence of instruc-
these definitions, (c) offer alternative tions because "the effects of the natu-
methods of describing the two sets of ral contingencies are known, therefore
behavioral data that evoke the verbal providing a baseline against which in-
response "sensitivity," and (d) discuss structional effects can be assessed" (p.
some of the implications of these pro- 465). One interpretation of this posi-
posed methods of description. tion is that researchers know what sen-
sitive performances look like: They
Definitions of Sensitivity look like schedule-typical nonhuman
and Insensitivity response patterns. Thus, when humans
fail to respond like nonhumans, the
Between-species comparisons. When performance is often described as be-
human behavior resembles schedule- ing insensitive to the programmed con-
typical nonhuman behavior maintained sequences.
by similar contingencies, researchers The B. Matthews et al. (1977) study
frequently describe the human behav- is not atypical. Much attention has
ior as being "sensitive" to the opera- been given to comparing human and
tive contingencies (Navarick, Bern- nonhuman patterns of behavior on
stein, & Fantino, 1990). Conversely, if schedules of reinforcement, and some
the human behavior is atypical of non- of the most influential work in this area
humans, the probability of labeling it has compared human schedule behav-
"insensitive" is increased (for a brief ior across age groups to nonhuman
summary of this practice in the human performances (e.g., Bentall, Lowe, &
operant literature, see Baxter & Schlin- Beasty, 1985; Lowe, Beasty, & Ben-
ger, 1990). We will refer to this relation tall, 1983). In these studies, preverbal
between human and nonhuman behav- infants' response patterns have most
ior as the between-species comparison resembled the scalloped and pause-re-
definition of sensitivity. For example, spond patterns viewed as typical of
Lowe, Harzem, and Bagshaw (1978) nonhumans on Fl schedules (although
suggested that observing-response see Hyten & Madden, 1993, for a re-
tasks may engender more sensitive hu- analysis of this data). The response
man fixed-interval (FI) performances patterns of older, language-able chil-
because the behavior obtained when dren in these studies have been more
these tasks are employed more closely comparable to adult human than to
 SENSITIVITY 3

nonhuman Fl-maintained behavior. measure of sensitivity. When the gen-

The impact of these studies on the de- eralized matching equation is applied
bate concerning human schedule sen- to concurrent-schedule performances,
sitivity is summarized by Baxter and the sensitivity parameter quantifies the
Schlinger (1990): "When schedule extent to which behavior changes with
sensitivity is assessed by comparisons changes in the relative frequency of re-
with nonhuman performances, only the inforcement. Sensitivity parameter val-
performances of preverbal children ues approximating zero indicate that
will resemble those of nonhumans" response allocation to the two rein-
(pp. 263-264). Such between-species forcement sources is affected little by
comparisons can be contrasted with the changes in the ratio of reinforcers ob-
other means of defining sensitivity: tained from the sources. So, for ex-
within-subject comparisons. ample, if the ratio of obtained reinforc-
Within-subject comparisons. The ers on the two schedules is initially 1:1
other class of behavioral data that leads and then changes to 5:1, but the re-
to using the term sensitivity is com-
posed of comparisons made within a sponse ratio remains at 1:1 throughout,
subject: The experimenter compares the exponent of the generalized match-
the behavior of a single subject in one ing equation will equal zero, indicating
condition with the behavior of the complete insensitivity to the change in
same subject under another condition. consequences. If, however, the re-
Behavior is described as being sensi- sponse ratio is initially 1:1 and then
tive when a stable pattern or rate of changes to 5:1 following the change in
behavior changes systematically fol- the ratio of obtained reinforcers, the re-
lowing a contingency change. Con- sponse ratio has changed proportion-
versely, stable behavior that fails to ally with the reinforcer change. The
change with the contingency change is latter case yields a sensitivity parame-
frequently described as being insensi- ter of 1.0 in the generalized matching
tive. We will refer to this definition as equation. If behavior were to change in
the within-subject comparison defini- the opposite direction (i.e., from a re-
tion of sensitivity. sponse ratio of 1:1 to 1:5) following
The human operant literature is re- the same contingency change, then be-
plete with examples of within-subject havior would likewise be sensitive to
comparisons in which behavior is de- this change and the sensitivity param-
scribed as being sensitive or insensitive eter would equal -1.0. The case of
to contingency changes (e.g., Baum, negative sensitivity parameter values,
1975; Buskist, Bennett, & Miller, although rarely observed, demonstrates
1981; Shimoff, Matthews, & Catania, that the sensitivity parameter quantifies
1986). Shimoff et al. (1986), for ex- behavior change following a change in
ample, described human performances contingencies of reinforcement, not
as insensitive to a contingency change changes in a particular direction. The
when response rates failed to change sensitivity parameter is not a measure
after the reinforcement contingency
was shifted from a tandem random-in- of the extent to which behavior resem-
terval (RI) differential-reinforcement- bles typical nonhuman performances
of-low-rate (DRL) schedule to a simple (nonhumans are usually undersensitive
RI schedule. with exponent values between 0.8 and
The within-subject comparison defi- 0.9; Baum, 1979; Wearden & Burgess,
nition of sensitivity has also been em- 1982) and therefore is inconsistent
ployed in several versions of Herrn- with the between-species comparison
stein's (1970) matching law. For ex- definition of sensitivity. This measure
ample, the sensitivity parameter of is a very precise example of the with-
Baum's (1974) generalized matching in-subject comparison definition of
equation is interpreted as a quantitative sensitivity, as specified above.
4 GREGORY J. MADDEN et al.
Confusions Created by Multiple there are other examples of data that
Definitions of Sensitivity can be interpreted as insensitive if
compared to nonhuman performances
These two uses of the term sensitiv- and sensitive if within-subject changes
ity are confusing because the same set in performances serve as criteria (e.g.,
of behavioral data can be viewed as be- Hayes, Brownstein, Zettle, Rosenfarb,
ing sensitive or insensitive depending & Korn, 1986).
on which definition is used. If a di-
mension of steady-state behavior (e.g., Confusions Created by
rate or postreinforcement pause dura- Between-Species Comparisons
tion) repeatedly covaries with dimen-
sions of the reinforcement schedules The procedural similarity of experi-
(e.g., mean interreinforcement inter- ments with nonhumans has led to a rel-
vals) and the experiment is apparently atively good understanding of the vari-
free of confounding variables, then ables that control nonhuman behavior.
contingency control has been demon- In addition, the procedures employed
strated and the behavior could be la- with nonhumans in separate laborato-
beled sensitive to the contingency ries are relatively consistent when
change. If the two steady-state perfor- compared with the array of procedures
mances, however, are unlike those of and reinforcers used with human sub-
nonhumans under similar contingen- jects (see the Spring, 1988, special is-
cies, researchers might conclude that sue of The Behavior Analyst). The
the behavior is insensitive to the op- standard procedures used with nonhu-
erative contingencies despite the obser- mans have resulted in a large literature
vation that it covaried with the contin- showing that behavior is sensitive to
gency change. parameters of reinforcement schedules.
Data reported by Weiner (1969) pro- These data supported researchers' gen-
vide an example of this confusion. In eralizations about nonhuman perfor-
Weiner's experiments, human subjects mances, like the Fl scallop, that were
given a history of reinforcement on a subsequently compared to human be-
DRL schedule produced response rates havior. Using the behavior of other
and postreinforcement pauses that co- species as a benchmark for human
varied with Fl schedule values (Exper- schedule sensitivity (between-species
iment 3). Thus, behavior of subjects comparison definition), however, is
with a DRL history was controlled by confusing for at least three reasons.
Fl contingencies, but the response pat- First, "schedule-typical" nonhuman
terns observed did not resemble those behavior may not be as typical as many
typical of nonhumans under similar suppose (Perone, Galizio, & Baron,
conditions; instead, one or two re- 1988). For example, Figure 1 shows
sponses were made at the end of each cumulative records from a single pi-
interval. If nonhuman response pat- geon under "sustained reinforcement"
terns serve as a benchmark of sensitiv- on an FH 1-min schedule (reprinted
ity against which human behavior is from Ferster & Skinner, 1957, p. 157).
compared, then this example of human The lower case letters are individual
behavior that was systematically af- intervals highlighted by Ferster and
fected by the contingency change Skinner across six sessions (A-F). Let-
would be described as being insensi- ters h, i, and k show break-and-run re-
tive to that change. If the within-sub- sponding like that found by Weiner
ject definition of sensitivity is applied (1969) with humans. The remaining
to Weiner's data, however, the post- lower case letters show intervals in
DRL behavior would be described as which the pigeon responded through-
being sensitive to the FH contingencies. out the Fl (other examples of this pat-
Weiner's data are not an isolated case. tern could be occasionally highlighted
Within the rule-governance literature, throughout session records D and E as
 SENSITIVITY 5

O0MINUTESA B C D EF

Figure 1. Cumulative records of pigeon behavior under an Fl I -min schedule of reinforcement

(reprinted from Ferster & Skinner, 1957, p. 157). The capital letters A through F show response
patterns from six individual sessions. The response patterns identified by lower case letters were
highlighted by Ferster and Skinner as exceptions to the scalloping otherwise observed. The original
figure caption read "Sustained reinforcement on FT 1."

well). Three other anomalous nonhu- or idealized pattern of nonhuman be-

man FH response patterns that were ob- havior as a benchmark against which
served frequently enough to warrant human behavior is judged to be sensi-
verbal description by Ferster and Skin- tive or insensitive to schedule contin-
ner include the "knee," a negative or gencies.
inverted scallop, and a single response Third, using the between-species def-
at the end of the interval. These anom- inition of sensitivity rests on the induc-
alous patterns bring into question how tive leap that procedures employed
characteristic of nonhuman FI behavior with humans and nonhumans that are
the scallop or break-and-run patterns structurally similar across species will
may be. be functionally similar as well. Accord-
Second, viewing nonhuman behav- ing to this logic, all experimental pro-
ior as a benchmark against which hu- cedures that resemble those controlling
man behavior is compared is confusing nonhuman behavior must also control
because behavior is not always entirely human behavior in the same manner.
consistent between nonhuman species. This conclusion, however, ignores the
When rats respond at low rates on possibility that procedures that are
DRL schedules and pigeons respond at structurally similar across experiments
higher rates (e.g., Kramer & Rilling, may produce functional differences
1970), which rate will researchers across species. Particular arrangements
compare with humans under similar of keys, reinforcer-delivery systems,
contingencies for the purpose of as- deprivation conditions, and so forth
sessing sensitivity? Similar differences have been used with particular species
separate different species under multi- because they are well suited to the phy-
ple, concurrent, and fixed-time sched- logenic histories of the species or to
ule contingencies, to name a few. Per- the ontogenic history of the organism.
one et al. (1988) have provided more When these same arrangements are
examples of interspecies performance used with another species, they may in-
differences in nonhumans. These ex- terfere with control by the variable of
amples argue against holding a typical interest (e.g., changes in schedules of
6 GREGORY J. MADDEN et al.

reinforcement). A number of studies man schedule sensitivity increased

that have focused on making human when procedures were employed that
operant procedures more functionally appear to more closely resemble the
similar to nonhuman procedures have functional characteristics of procedures
found human behavior to be highly used with nonhumans.
sensitive to changing parameters of re- As noted above, some researchers
inforcement (e.g., Barnes & Keenan, have apparently assumed that the con-
1993; Baron & Kaufman, 1966; Baum, tingency changes employed in human
1975; Baxter & Schlinger, 1990; Busk- and nonhuman operant experiments are
ist & Miller, 1981; Galizio, 1979; Hy- functionally similar, and this has led
ten, Madden, & Field, 1994; Joyce & them to suggest that instances of hu-
Chase, 1990; LeFrancois, Chase, & man schedule insensitivity call for new
Joyce, 1988; Logue, Forzano, & Tobin, principles of behavior (e.g., Home &
1992; Madden & Perone, in press; Lowe, 1993; Lowe, 1979). An analo-
Schroeder & Holland, 1969; Torgrud & gous instance of interspecies sensitivity
Holburn, 1990; Trenholme & Baron, differences currently exists in the non-
1975). In addition, other studies have human concurrent-schedule literature,
found nonhuman behavior to more but the scientific community's reaction
closely resemble human behavior has been rather different. Briefly, most
when procedures were made to func- nonhuman concurrent VI VI behavior
tionally resemble those frequently em- is well described by the matching law
ployed with human subjects (e.g., (Herrnstein, 1970) with slight under-
Jackson & Hackenberg, 1996; Wan- matching (for reviews see Bradshaw &
chisen, Tatham, & Mooney, 1989). Szabadi, 1988; de Villiers, 1977).
The argument that human and non- Cows, however, are much less sensitive
human behavior is governed by differ- to concurrent schedules, in a manner
ent processes has most recently fo- not unlike human behavior under the
cused on human sensitivity to concur- same schedules (Foster, Temple, Rob-
rent variable-interval (VI) VI sched- ertson, Nair, & Poling, 1996; L. Mat-
ules (Home & Lowe, 1993). Although thews & Temple, 1979). Interestingly,
Home and Lowe have described con- no researchers have claimed (to our
ditions under which humans are insen- knowledge) that new principles of be-
sitive to concurrent schedules, an ex- havior are required to understand the
periment conducted by Madden and behavior of cows, or that processes
Perone (in press) suggests that this in- governing cow behavior also govern
sensitivity may be the result of func- human behavior. Instead, efforts to un-
tional differences in procedures sepa- derstand sensitivity differences be-
rating experiments with humans and tween cows and other nonhumans have
nonhumans. Arguing that procedures been focused primarily on the proce-
typically employed with humans do dural differences separating these ex-
not require subjects to observe the periments (e.g., Dougherty & Lewis,
stimuli correlated with the concurrent- 1992; Foster et al., 1996; Rachlin, Ka-
schedule alternatives, Madden and Per- gel, & Battalio, 1980). We believe that
one manipulated the extent to which the same strategy may be well suited
human subjects could make an observ- to understanding instances of human
ing response to produce these stimuli. insensitivity to schedules of reinforce-
Behavior was insensitive when sub- ment.
jects either could not or did not ob-
serve the schedule-correlated stimuli. Toward a Definition of Sensitivity
Schedule sensitivity increased within
the same subjects, however, when pro- As a first step toward a better defi-
cedural changes were introduced that nition of sensitivity, we propose that
required the subjects to observe the sensitivity is not determined by com-
stimuli to earn reinforcers. Thus, hu- paring human and nonhuman behavior.
 SENSITIVITY 7

Our previous arguments suggest that careful attention to controlling poten-

schedule-typical behavior is difficult to tial confounding variables, can contrib-
identify, and that there are many func- ute to our knowledge of procedural or
tional and structural differences in pro- subject variables that might contribute
cedures used with different species. By to insensitivity. The intent here, there-
arguing against the use of nonhuman fore, is not to criticize experiments us-
benchmarks for determining sensitivi- ing between-subjects comparison pro-
ty, however, we are not suggesting that cedures; instead our focus is on the use
researchers should ignore interspecies of between-species comparisons as the
differences in behavior. Overlooking basis for statements about human sen-
such differences may lead to inaccurate sitivity and insensitivity to experimen-
generalizations or predictions and to tal variables.
stifled research in areas mistakenly Having parceled out the between-
considered to be well understood. We species definition of sensitivity, we are
think the term interspecies replication left with the following: Sensitivity is
better describes the consistency of ef- demonstrated when an experimental
fect obtained across species than does manipulation affects behavior in an or-
the term sensitivity. The former term derly and replicable manner. Insensitiv-
clearly specifies the behavioral data ity describes a lack of behavior change
that evoke its use, leaving sensitivity to following an experimental manipula-
describe effects of the independent tion. As noted above, this within-sub-
variable on individual behavior (Baxter ject definition of sensitivity is consis-
& Schlinger, 1990). Thus, Weiner tent with the definition of sensitivity
(1969) provided data that showed hu- within Baum's (1974) generalized
man sensitivity to the manipulated matching equation. In fact, the sensi-
schedules but failed to demonstrate in- tivity parameter within the matching
terspecies replication. law provides a precise quantitative
Our discussion of the between-spe- measure of the degree of sensitivity.
cies definition of human schedule sen- Another quantitative measure consis-
sitivity should not be viewed as an in- tent with the within-subject definition
dictment of between-subjects compar- of sensitivity is elasticity of demand
isons for investigating the effects of (Hursh, 1980). Elasticity measures sen-
experimental variables. Comparisons sitivity of consumer demand (i.e., the
made between 2 human subjects ex- rate at which reinforcers are obtained)
posed to different conditions (e.g., B. to changing reinforcement magnitude
Matthews et al., 1977) or differing in or schedule parameters. In both cases,
subject characteristics (e.g., Bentall et a form of experimental control is re-
al., 1985) but otherwise exposed to quired in order to use the terms sensi-
identical schedules of reinforcement tivity or elasticity. Thus, sensitive be-
are frequently employed in the human havior is synonymous with behavior
operant literature. This is particularly under experimental control.
true in the rule-governance literature, Two experiments provide examples
in which subjects are frequently given of the above definitions of sensitivity,
different instructions and exposed to insensitivity, and interspecies replica-
identical schedules of reinforcement. tion. First, Baxter and Schlinger (1990)
Although within-subject procedures of- reported instances of sensitivity and in-
fer the experimenter more power to de- terspecies replication in their study of
tect the effects of independent vari- children's response rates under ran-
ables, these between-subjects compar- dom-ratio (RR) and RI contingencies.
isons may be necessary either because Behavior was sensitive to the schedule
providing instructions cannot be re- type because response rates were reli-
versed or because subject variables ably higher under the RR schedule.
cannot be manipulated within a sub- The same data provide evidence for in-
ject. These comparisons, if done with terspecies replication because nonhu-
8 GREGORY J. MADDEN et al.

mans also respond more rapidly on RR failure to obtain interspecies replica-

than RI schedules (e.g., Catania, Mat- tion. This difference, although it does
thews, Silverman, & Yohalem, 1977). not show sensitivity, strongly suggests
Pierce, Epling, and Greer (1981) careful consideration of the similarity
conducted an experiment in which hu- of the procedures used across species.
man speech directed toward two con- For example, in the Pierce et al. ex-
federates was verbally praised accord- periment in which negative sensitivity
ing to concurrent VI VI schedules. The parameter values were observed, the
first subject was insensitive to changes investigators arguably used qualitative-
in the relative rate of praise given by ly different reinforcers by having so-
each confederate (sensitivity parameter cial praise delivered by two different
of the generalized matching equation, confederates. In animal experiments,
a = -0.02; where a indicates greater reinforcers obtained from both concur-
sensitivity as values deviate further rent schedules are typically of the same
from zero in either a positive or nega- type (e.g., grain) and come from the
tive direction). That is, the first sub- same location (e.g., a single food hop-
ject's speech was about equally distrib- per).
uted between confederates, despite one Some readers may object to labeling
delivering more praise than another. a negative sensitivity parameter value
Behavior of the third subject showed as an instance of concurrent-schedule
greater sensitivity than any other sub- sensitivity because this behavior seems
ject in the experiment (a = -0.49); "irrational" in the economic sense of
however, sensitivity parameter values the word. From our view, however, ir-
are typically positive in nonhumans. rationality has nothing to do with sen-
Thus, interspecies replication was not sitivity. Behavior that changes system-
demonstrated by either of these sub- atically with changes in an independent
jects. variable suggests that a controlling
variable has been isolated. If sensitive
Implications of the Proposed behavior also seems irrational, then the
Definitions logic behind the rational explanation of
the manipulated variables needs to be
Perhaps one obvious implication of examined. In the Pierce et al. (1981)
the proposed definition of sensitivity is experiment, for example, the possibil-
that the term should not be viewed as ity exists that the negative sensitivity
being synonymous with some logically parameter value is due to the existence
defined schedule-appropriate respond- of a complementary relation between
ing, matching, or maximizing. For ex- the two reinforcers (e.g., Allison, 1983;
ample, the negative sensitivity param- Bickel, DeGrandpre, & Higgins, 1995;
eter value reported by Pierce et al. Rachlin et al., 1980). When a comple-
(1981) is not an example of insensitiv- mentary relation exists between two
ity, even though the negative value in- different reinforcers (e.g., between
dicates that more behavior was allo- food and water), the reinforcers lose
cated to the relatively leaner concur- some of their efficacy if they are not
rent-schedule alternative. Although this obtained at a constant ratio (e.g., two
subject obviously failed to maximize units of food to every one unit of water
reinforcers while minimizing response consumed). The subject in question in
output, the behavior was systematically the Pierce et al. experiment may have
affected by the different contingency allocated more behavior to the relative-
changes imposed in the experiment. ly leaner schedule of reinforcement be-
Because nonhumans nearly always al- cause deviating from a constant ratio
locate their behavior in the exact op- of reinforcers obtained from the two
posite fashion (i.e., they allocate more confederates (e.g., 1:1) would have re-
behavior to the relatively richer avail- sulted in a loss of reinforcer efficacy.
able schedule), this is an example of These speculations are not intended to
 SENSITIVITY 9
serve as a definitive analysis, but in- by the contingencies of doing science,
stead simply point out that despite their and all may be necessary in order to
irrationality, any instance of behavioral find what works.
sensitivity is an interesting phenome- The final implication of the pro-
non worthy of further study (e.g., Skin- posed definitions is that the difficulties
ner, 1956). associated with using between-species
A second implication of the pro- comparisons to determine schedule
posed definitions is that instances of sensitivity frequently appear equally
human insensitivity to changing sched- applicable to comparisons made be-
ules of reinforcement may have theo- tween human subjects of different
retical significance if the experiments stages of development (e.g., Bentall et
that verify this insensitivity have ma- al., 1985; Darcheville, Riviere, &
nipulated the full parametric range of Wearden, 1993). That is, differences
the variable of interest, and if nonhu- that may be attributed to developmen-
man behavior is sensitive to these tal differences separating subject
changes (note that we are interested in groups might also be attributable to
interspecies replicability of sensitivity, procedural differences separating the
rather than human and nonhuman be- groups. For example, in the Bentall et
havior being identical). Before the in- al. (1985) study, differences in FH
vestigator may assert a discontinuity schedule-maintained response patterns
between human and nonhuman behav- observed across the infant and older
ior, however, discontinuities between developmental groups may have been
the procedures employed with humans a function of developmental differ-
and nonhumans must be experimental- ences (e.g., language capacity), but
ly investigated. The list of potentially they may have also been due to differ-
important procedural differences is ences in experimental settings, manip-
likely to be formidable and represents ulanda, and reinforcers. Whether these
a challenging task for the scientist who structural differences in procedure
seeks to understand interspecies dis- amount to functional differences af-
continuity. Indeed, the list of possible fecting behavior is an empirical ques-
procedural differences is huge, and a tion requiring further experimental
researcher biased toward seeing inter- analysis; the current data do not ade-
species continuity may be able to cre- quately support either position.
ate a seemingly endless list of these
variables, just as a researcher biased Conclusions
toward identifying instances of discon-
tinuity may be able to provide an end- Few issues in the experimental anal-
less list of reasons discounting the va- ysis of behavior are more important
lidity of these procedural concerns. De- than the generality of operant princi-
finitive demonstration of interspecies ples to the behavior of humans. Be-
continuity or lack of continuity may be cause the adoption and use of operant
an impossible task, rendering the po- contingencies to change human behav-
sitions as opposing philosophical as- ior hinge upon successful demonstra-
sumptions. tions of human contingency sensitivity
We are not interested in limiting the both in and outside the laboratory, it is
debate fostered by such opposing po- critical that agreed-upon definitions of
sitions. The behavior of scientists, like sensitivity and insensitivity are applied
the behavior of members of all groups, to these data. We have argued that a
can be viewed in terms of its variabil- criterion of sensitivity based on be-
ity. There are those who take extreme tween-species comparisons has led to a
positions on each end of any issue and number of confusions; perhaps most
those who take moderate positions be- notably that unexplored functional dif-
tween these ends. We suspect all of ferences may separate structurally sim-
these various positions are supported ilar human and nonhuman experimen-
 10 GREGORY J. MADDEN et al.

tal procedures, and that this unexplored performance. Journal of the Experimental
status renders premature many conclu- Analysis of Behavior, 59, 501-520.
Baron, A., & Galizio, M. (1983). Instructional
sions about the generality (or the lack control of human operant behavior. The Psy-
of generality) of the processes govern- chological Record, 33, 495-520.
ing nonhuman behavior to the behavior Baron, A., & Kaufman, A. (1966). Human free-
of humans. Thus, we recommend using operant avoidance of "time out" from mon-
etary reinforcement. Journal of the Experi-
a within-subject definition of schedule mental Analysis of Behavior, 9, 557-565.
sensitivity, and separately describing Baum, W. M. (1974). On two types of deviation
the extent to which interspecies repli- from the matching law: Bias and undermatch-
cations are demonstrated. ing. Journal of the Experimental Analysis of
Recognizing that insensitivity is Behavior, 22, 231-242.
Baum, W. M. (1975). Time allocation in human
synonymous with failure of experimen- vigilance. Journal of the Experimental Anal-
tal control suggests that when there is ysis of Behavior, 23, 45-53.
reason to believe that the behavior of Baum, W. M. (1979). Matching, undermatch-
nonhuman organisms is sensitive to a ing, and overmatching in studies of choice.
Journal of the Experimental Analysis of Be-
particular independent variable and havior, 32, 269-281.
within-session manipulations reveal Baxter, G. A., & Schlinger, H. (1990). Perfor-
that human behavior is insensitive to mance of children under a multiple random-
this variable, then in these cases the ex- ratio random-interval schedule of reinforce-
ment. Journal of the Experimental Analysis of
perimenter should focus on procedural Behavior, 54, 263-271.
factors that may have led to insensitiv- Bentall, R. P., Lowe, C. F, & Beasty, A. (1985).
ity. Only after these factors have been The role of verbal behavior in human learn-
ruled out would a finding of insensitiv- ing: II. Developmental differences. Journal of
the Experimental Analysis of Behavior, 43,
ity be interpreted as having theoretical 165-181.
significance, and then only within a Bickel, W. K., DeGrandpre, R. J., & Higgins, S.
carefully argued network of findings. T. (1995). The behavioral economics of con-
Although we recommend within- current drug reinforcers: A review and reanal-
subject experiments for assessing sen- ysis of drug self-administration research. Psy-
chopharmacology, 118, 250-259.
sitivity to the effects of an independent Bradshaw, C. M., & Szabadi, E. (1988). Quan-
variable, we recognize the importance titative analysis of human operant behavior. In
of between-species and between-sub- G. Davey & C. Cullen (Eds.), Human operant
jects comparisons. Between-species conditioning and behavior modification (pp.
225-259). Chichester, England: Wiley.
comparisons are important in assessing Brewer, W. F. (1974). There is not convincing
the interspecies generality of behavior- evidence for operant or classical conditioning
al principles and in understanding how in adult humans. In W. B. Weimer & D. J.
phylogenic differences affect the be- Palermo (Eds.), Cognition and symbolic pro-
cesses (pp. 1-42). Hillsdale, NJ: Erlbaum.
havior of different species. Likewise, Buskist, W. F, Bennett, R. H., & Miller, H. L.
between-subjects comparisons are im- (1981). Effects of instructional constraints on
portant in assessing the effects of sub- human fixed-interval performance. Journal of
ject variables and variables that are the Experimental Analysis of Behavior, 35,
suspected of having irreversible ef- 217-225.
Buskist, W. F, & Miller, H. L. (1981). Concur-
fects. When using these methodolo- rent operant performance in humans: Match-
gies, however, the researcher must be ing when food is the reinforcer. The Psycho-
careful to avoid making statements logical Record, 31, 95-100.
about the sensitivity or insensitivity of Catania, A. C., Matthews, B. A., Silverman, P.
J., & Yohalem, R. (1977). Yoked variable-
a particular species' or subject's behav- ratio and variable-interval responding in pi-
ior based on these comparisons alone. geons. Journal of the Experimental Analysis
of Behavior, 28, 155-161.
REFERENCES Catania, A. C., Shimoff, E., & Matthews, B. A.
(1989). An experimental analysis of rule-gov-
emed behavior. In S. C. Hayes (Ed.), Rule-
Allison, J. (1983). Behavioral economics. New governed behavior: Cognition, contingencies,
York: Praeger. and instructional control, (pp. 119-150). New
Barnes, D., & Keenan, M. (1993). Concurrent York: Plenum Press.
activities and instructed human fixed-interval Cerutti, D. T. (1989). Discrimination theory of
 SENSITIVITY 11
rule-governed behavior. Journal of the Exper- behavior. Psychonomic Monograph Supple-
imental Analysis of Behavior, 51, 259-276. ments, 1, 243-250.
Darcheville, J. C., Riviere, V., & Wearden, J. H. Kramer, T. J., & Rilling, M. (1970). Differen-
(1993). Fixed-interval performance and self- tial-reinforcement-of-low-rates: A selective
control in infants. Journal of the Experimental critique. Psychological Bulletin, 74, 225-256.
Analysis of Behavior, 60, 239-254. LeFrancois, J. R., Chase, P. N., & Joyce, J. H.
de Villiers, P. A. (1977). Choice in concurrent (1988). The effects of a variety of instructions
schedules and a quantitative formulation of on human fixed-interval performance. Journal
the law of effect. In W. K. Honig & J. E. R. of the Experimental Analysis of Behavior, 49,
Staddon (Eds.), Handbook of operant behav- 383-393.
ior (pp. 233-287). Englewood Cliffs, NJ: Logue, A. W., Forzano, L. B., & Tobin, H.
Prentice Hall. (1992). Independence of reinforcer amount
Dougherty, D. M., & Lewis, P. (1992). Match- and delay: The generalized matching law and
ing by horses on several concurrent variable- self-control in humans. Learning and Moti-
interval schedules. Behavioural Processes, 26, vation, 23, 326-342.
69-76. Lowe, C. F (1979). Determinants of human op-
Ferster, C. B., & Skinner, B. F. (1957). Sched- erant behaviour. In M. D. Zeiler & P. Harzem
ules of reinforcement. New York: Appleton- (Eds.), Reinforcement and the organization of
Century-Crofts. behavior (pp. 159-192). New York: Wiley.
Foster, T. M., Temple, W., Robertson, B., Nair, Lowe, C. F, Beasty, A., & Bentall, R. P. (1983).
V., & Poling A. (1996). Concurrent-schedule The role of verbal behavior in human learn-
performance in dairy cows: Persistent under- ing: Infant performance on fixed-interval
matching. Journal of the Experimental Anal- schedules. Journal of the Experimental Anal-
ysis of Behavior, 65, 57-80. ysis of Behavior, 39, 157-164.
Galizio, M. (1979). Contingency-shaped and Lowe, C. F, Harzem, P., & Bagshaw, M. (1978).
rule-governed behavior: Instructional control Species differences in temporal control of be-
of human loss avoidance. Journal of the Ex- havior II: Human performance. Journal of the
perimental Analysis of Behavior, 31, 53-70. Experimental Analysis of Behavior, 29, 351-
Hayes, S. C., Brownstein, A. J., Zettle, R. D., 361.
Rosenfarb, I., & Korn, Z. (1986). Rule-gov- Lowe, C. F, & Home, P J. (1996). Reflections
erned behavior and sensitivity to changing on naming and other symbolic behavior. Jour-
consequences of responding. Journal of the nal of the Experimental Analysis of Behavior,
Experimental Analysis of Behavior, 45, 237- 65, 315-340.
256. Madden, G. J., & Perone, M. (in press). Human
Herrnstein, R. J. (1970). On the law of effect. sensitivity and insensitivity to concurrent
Journal of the Experimental Analysis of Be- schedules of reinforcement: Effects of observ-
havior, 13, 243-266. ing schedule-correlated stimuli. Journal of the
Horne, P J., & Lowe, C. F (1993). Determi- Experimental Analysis of Behavior.
nants of human performance on concurrent Matthews, B. A., Shimoff, E., Catania, A. C., &
schedules. Journal of the Experimental Anal- Sagvolden, T. (1977). Uninstructed human
ysis of Behavior, 59, 29-60. responding: Sensitivity to ratio and interval
Horne, P J., & Lowe, C. F (1996). On the or- contingencies. Journal of the Experimental
igins of naming and other symbolic behavior. Analysis of Behavior, 27, 453-467.
Journal of the Experimental Analysis of Be- Matthews, L. R., & Temple, W (1979). Con-
havior, 65, 185-241. current schedule assessment of food prefer-
Hursh, S. R. (1980). Economic concepts for the ence in cows. Journal of the Experimental
analysis of behavior. Journal of the Experi- Analysis of Behavior, 32, 245-254.
mental Analysis of Behavior, 34, 219-238.
Hyten, C., & Madden, G. J. (1993). The scallop Navarick, D. J., Bernstein, D. J., & Fantino, E.
in human fixed-interval research: A review of (1990). The experimental analysis of human
problems with data description. The Psycho- behavior. Journal of the Experimental Analy-
logical Record, 43, 471-500. sis of Behavior, 54, 159-162.
Hyten, C., Madden, G. J., & Field, D. P. (1994). Perone, M., Galizio, M., & Baron, A. (1988).
Exchange delays and impulsive choice in The relevance of animal-based principles in
adult humans. Journal of the Experimental the laboratory study of human operant con-
Analysis of Behavior, 62, 225-233. ditioning. In G. Davey & C. Cullen (Eds.),
Jackson, K., & Hackenberg, T. D. (1996). To- Human operant conditioning and behavior
ken reinforcement, choice, and self-control in modification (pp. 59-85). Chichester, En-
pigeons. Journal of the Experimental Analysis gland: Wiley.
of Behavior, 66, 29-49. Pierce, W D., Epling, W. F, & Greer, S. M.
Joyce, J. H., & Chase, P N. (1990). Effects of (1981). Human communication and the
response variability on the sensitivity of rule- matching law. In C. M. Bradshaw, E. Szabadi,
governed behavior. Journal of the Experimen- & C. F Lowe (Eds.), Quantification ofsteady-
tal Analysis of Behavior, 54, 251-262. state operant behavior (pp. 345-348). Am-
Kaufman, A., Baron, A., & Kopp, R. E. (1966). sterdam: Elsevier.
Some effects of instructions on human operant Rachlin, H., Kagel, J. H., & Battalio, R. (1980).
12 GREGORY J. MADDEN et al.
Substitutability in time-allocation. Psycholog- Torgrud, L. J., & Holburn, S. W. (1990). The
ical Bulletin, 87, 355-374. effects of verbal performance descriptions on
Schroeder, S. R., & Holland, J. G. (1969). Re- nonverbal operant responding. Journal of the
inforcement of eye movement with concurrent Experimental Analysis of Behavior, 54, 273-
schedules of reinforcement. Journal of the Ex- 291.
perimental Analysis of Behavior, 12, 897-903. Trenholme, I. A., & Baron, A. (1975). Imme-
Shimoff, E., Catania, A. C., & Matthews, B. A. diate and delayed punishment of human be-
(1981). Uninstructed human responding: Sen- havior by loss of reinforcement. Learning and
sitivity of low-rate performances to schedule Motivation, 6, 62-79.
contingencies. Journal of the Experimental Wanchisen, B. A., Tatham, T. A., & Mooney, S.
Analysis of Behavior, 36, 207-220. E. (1989). Variable-ratio conditioning history
Shimoff, E., Matthews, B. A., & Catania, A. C. produces high- and low-rate fixed-interval
(1986). Human operant performance: Sensi- performance in rats. Journal of the Experi-
tivity and pseudosensitivity to contingencies. mental Analysis of Behavior, 52, 167-179.
Journal of the Experimental Analysis of Be-
havior, 46, 149-157. Wearden, J. H., & Burgess, I. S. (1982). Match-
Skinner, B. F (1956). A case history in scien- ing since Baum (1979). Journal of the Ex-
tific method. American Psychologist, 11, 211- perimental Analysis of Behavior, 38, 339-
233. 348.
Skinner, B. F (1966). An operant analysis of Weiner, H. (1969). Controlling human fixed-in-
problem-solving. In B. Kleinmuntz (Ed.), terval performance. Journal of the Experimen-
Problem solving (pp. 225-257). New York: tal Analysis of Behavior, 12, 349-373.
Wiley.