Tutorial Stimulus Control: Part II James A

The Behavior Analyst 1995, 18, 253-269 No. 2 (Fall) Tutorial Stimulus Control: Part II James A. Dinsmoor Indiana University

The second part of my tutorial stresses the systematic importance of two parameters of discrimination training: (a) the magnitude of the physical difference between the positive and the negative stimulus (disparity) and (b) the magnitude of the difference between the positive stimulus, in particular, and the background stimulation (salience). It then examines the role these variables play in such complex phenomena as blocking and overshadowing, progressive discrimination training, and the transfer of control by fading. It concludes by considering concept formation and imitation, which are important forms of application, and recent work on equivalence relations. Key words: stimulus control, disparity, salience, blocking, overshadowing, transfer, fading, concept formation, imitation, equivalence relations

The first part of this tutorial dealt a positive correlation between the S + with the basic principles that account and the primary reinforcer that selects for the acquisition of stimulus control the one relevant stimulus out of the under what is conventionally known as many that impinge upon the organism discrimination training. Among the sa- and transforms it into a conditioned re- lient points that were raised were sug- inforcer of observing behavior. Note gestions that (a) control by antecedent also that if the subject is exposed ex- stimuli is just as important in operant clusively to S + training or to a single as it is in respondent behavior; (b) Pav- period of S + training followed by a lov's conditional stimulus is a discrim- single period of S - training rather than inative stimulus; (c) stimulus general- to a continued alternation, control is ization is not a behavioral process in- less adequate (Honig, Thomas, & Gutt- dependent of and antagonistic to dis- man, 1959; Yarczower & Switalski, crimination but is simply another way 1969). Similarly, in a compound dis- of describing control by antecedent crimination like that studied by Blough stimuli; and (d) the increases in control (1969), if one dimension is left contin- that occur during discrimination train- uously at its positive value for a num- ing can be attributed to more frequent ber of sessions, control by that dimen- and more prolonged observation of the sion is reduced and control by the other relevant stimuli accompanied, it is pre- dimension enhanced; when the proce- sumed, by concomitant changes in at- dure returns to the previous alternation tention. between positive and negative stimuli, Note that in conventional discrimi- the discriminative performance returns nation training, the subject is exposed to its normal level. to repeated alternations between the The second part of my tutorial will positive stimulus (S+), which is ac- necessarily be less tightly integrated. companied by a schedule of reinforce- Taking off from the foundation laid in ment, and an alternative stimulus (S-), Part I, it extends the treatment of stim- which is not accompanied by rein- ulus control to a discussion of two of forcement. This alternation establishes its most important parameters and to several more complex patterns that Address correspondence to James A. Dins- seem to hold special significance for moor, Department of Psychology, Indiana Uni- basic theorizing and practical applica- versity, Bloomington, Indiana 47405. tion. 253 254 JAMES A. DINSMOOR

STIMULUS PARAMETERS er intensity was set at 60 dB, with the higher intensity determined by the There are a number of variables that magnitude of the difference between influence the rate at which the subject the two; for the other half, the higher learns to discriminate between stimuli intensity was set at 100 dB, with the and the level of performance that will lower intensity determined by the mag- ultimately be attained. However, some nitude of the difference. (Because there of these, like the physical quality of the was only one 60-100 group and one stimulus (e.g., wavelength of light, en- 100-60 group, there were 14 rather ergy of sound), the topography of the than 16 groups in all, plus a special response, and individual and species control group.) differences among different subjects, It seems obvious that very small do not readily lend themselves to the physical differences between the posi- formulation of general laws of behav- tive stimulus (SD or S+) and the neg- ior. Of greater interest from a system- ative stimulus (S^ or S-) must be dif- atic point of view are those parameters ficult to discriminate. Up to some limit, that enter into a variety of behavioral at least, larger differences should pro- paradigms. There are two parameters, mote faster acquisition and a larger ul- in particular, that will appear and reap- timate difference between the two per- pear in subsequent accounts of topics formances. This expectation is borne like overshadowing, blocking, the out by a series of plots tracing the easy-to-hard effect, fading, and con- course of the discrimination index cept formation. These are (a) the mag- (multiplied by 10) as a function of the nitude of the difference in physical hours of training. In all four panels of units between the positive and the neg- Figure 1, the discrimination is slowest ative stimulus, sometimes subsumed to develop and attains the lowest final by the phrase stimulus disparity, and level in groups for which the stimuli (b) the magnitude of the difference be- differ by only 10 dB; groups for which tween the discriminative stimuli and the difference is 20 dB do somewhat the background stimulation, subsumed better; groups for which the difference as stimulus salience. is 30 or 40 dB differ less during acquisition and tend to converge, but at Stimulus Disparity still higher levels of performance. With Clearcut evidence for the role played pigeons as subjects, Hanson (1959) by the disparity between the two stim- found a similar relation between the uli may be found in a series of early magnitude of the difference in wave- studies, using rats as subjects, con- length and the time required for the de- ducted by Rosemary Pierrel and her as- velopment of a discrimination. Again sociates at Brown University. Bar the slope of the function was steep at pressing was reinforced on a variable- small values but decreased as larger interval schedule in the presence of the values were approached. positive stimulus but not in the presence of the negative stimulus. In Pier- Stimulus Salience rel, Sherman, Blue, and Hegge (1970), for example, the discriminative stimuli Further examples of the effects of were pulsed tones of 4 kHz that dif- stimulus disparity would not be hard to fered in intensity. The difference be- find, but the magnitude of the differ- tween the positive and the negative ence between the discriminative stim- stimulus was set at 10, 20, 30, or 40 uli and their background stimulation dB, with the higher intensity serving as (salience) poses more of a problem. the positive stimulus for half the This is a dimension that does not come groups and as the negative stimulus for up for consideration within the older the other half. Also, for half the ani- response-strengthening-and-weakening mals, independently assigned, the low- theories of discrimination learning. STIMULUS CONTROL 255 A B

900 100

[V~~ ~ ~ ~ ~ ~ ~ ~ ~ _

700/ :! ~~60- 707!-6

X600 60 -80 ---- 80-m60-- F Pct60 100 - 60 z o 500 P

o900 V , ~ ~ OR IN LCSO

SD ofDBA 100- 90- a90 - 00- 600 loo~0- 80-- 80 - 100--- spcfew ndcbl.(erdcdfo ire ta. 90 cprgh Socet fo th xermna 100 - 70-''-70 - tO .....- 100-~~~~~~~~606- 100- 500

32 96 160 224 32 96 160 224 HOURS IN BLOCKS OF 8 Figure 1. Percentage of responding in SD (S+), multiplied by 10, as a function of hours of training. Each point represents the mean of 4 rats during an 8-hr session. The stimuli for each group are specified in decibels. (Reproduced from Pierrel et al., 1970; copyright Society for the Experimental Analysis of Behavior.)

However, when investigators are guid- stimulus disparity on the relative rates ed by an interest in the role of observ- of pecking two observing keys (Dins- ing or attention, it becomes a more moor, Sears, & Dout, 1976). In our first likely candidate. In my laboratory, we experiment, the birds pecked a key that became aware of the importance of sa- produced large increases or decreases lience while studying the effects of in illumination as discriminative stim- 256 JAMES A. DINSMOOR

uli at higher rates than they pecked a by depressing a low-lying "perch," key that produced smaller increases or similar to the conventional cross-bar decreases in illumination. In our sec- for rats, the pigeon produced stimuli ond experiment, however, we varied that were correlated with the compo- the magnitude of change for the posi- nent schedules. During periods when tive and the negative stimuli indepen- the variable-interval schedule was in dently. We discovered that the positive effect, the stimulus was, for two dif- relation between magnitude of change ferent groups, a clockwise tilt of 150 or and rate of pecking stemmed entirely a clockwise tilt of 300 (S+); during pe- from the positive stimulus: For the riods when the grain was being with- negative stimulus, the function was held, it was a counterclockwise tilt of negative in slope. That is, larger the same magnitude (S-). For half of changes in stimulation reduced the the subjects, then, the total difference rate. On returning to the experimental between the two tilts was 30°, and for literature, we were led to the conclu- the other half, it was 600. These differ- sion that in most previous examina- ences represented small and large dis- tions of stimulus difference, the mag- parities, respectively, between the two nitude of the difference between the stimuli. two stimuli had been confounded with To vary the salience of the stimuli, the magnitude of the difference be- the experimenters used an arrangement tween the discriminative stimuli and similar to that of Johnson (1970), re- their background. That is, the disparity ducing the contrast between the black betwen the stimuli had been confound- line, regardless of its tilt, and the white ed with their salience. surround. This was accomplished by At the time, we were not aware that the simultaneous lighting of two pro- a study of discrimination in which the jection units, one of which displayed salience of the stimuli had indeed been an image of the line on the key and the varied independently of other factors other of which produced only a uni- had already been conducted by John- form white field. Both lamps were re- son (1970). A student of Cumming, duced to 50% of their normal intensity. Johnson was interested in the role of Thus, the salience of the stimuli could attention in discrimination learning, be manipulated without affecting their and one of the parameters he had var- disparity. When the image of the line ied, as part of a larger study of selec- was projected at its maximal contrast tive control, was the brightness of a by a single projector cell, 7 of the 8 white line displayed on the pigeon's birds in that group learned both to hold key. Pecks when the line was vertical down the perch and to discriminate in (S+) were reinforced on a random-in- their rate of pecking between the re- terval schedule, but pecks when the sulting stimuli. When the second cell line was horizontal (S-) were never was lighted, the line, now gray, was reinforced. The rate at which the birds readily visible to the human eye and learned the discriminaton between presumably to that of the pigeon. Nev- these two orientations was a function ertheless, only 1 of the 8 birds in the of the brightness of the line. other group gave any indication that its Still blithely unaware of Johnson's behavior was influenced by the tilt of study, Dinsmoor, Mueller, Martin, and the line. Obviously, the salience of the Bowe (1982) chose as their controlling stimuli was an important determinant stimulus a black line bisecting a white of the level of observing and of the key. Pecks on the key produced grain consequent level of discrimination. on a variable-interval schedule that al- The effects of disparity were not as ternated with an extinction schedule. clear as those for salience, but in a sub- Ordinarily, the line remained horizon- sequent study, using an entirely differ- tal in its alignment, regardless of which ent design, Dinsmoor et al. (1983) of these schedules was operating. But found that both of these dimensions af- STIMULUS CONTROL 257 fected the rate at which pigeons pecked the same year, a set of experiments an observing key. published by Wagner, Logan, Haber- landt, and Price (1968) implicated yet SELECTIVE CONTROL a third variable, the consistency with which each member of the stimulus It was Pavlov (1927/1960, pp. compound was followed by the uncon- 141ff.) who first called attention to the ditional stimulus (US). A CS that was reduction in the effectiveness of a stim- always followed by the US reduced the ulus that sometimes occurred when effectiveness of a stimulus that was that stimulus was presented in tandem followed by the US on only half of its with another stimulus. He found that presentations. when two conditional stimuli were reg- The common thread running through ularly presented together, in the same the loss of effectiveness in each of temporal relation to each presentation these instances is that both stimuli are of the unconditional stimulus, one regularly presented at the same time member of the pair often came to as- and therefore in the same temporal lo- sume complete control over the cus with regard to the unconditional amount of saliva that was secreted, to stimulus. Other ways of describing the the partial or complete exclusion of the relation between the two CSs are to say other stimulus. That is, when tested that they covary, that they are con- alone, the second stimulus was ineffec- founded, or that they duplicate one an- tive. On further examination, he found other. Cognitive psychologists often re- that it was the relative intensity of the fer to the second stimulus as "redun- two conditional stimuli (CSs) that de- dant," meaning that it provides no ad- termined which of them prevailed and ditional information concerning the the extent to which the effectiveness of arrival of the US (i.e., no improvement the less intense stimulus was reduced. in the prediction of the US over that When the stimuli were of equal inten- provided by the first stimulus). Both sity (e.g., two tones of equal loudness), stimuli convey the same meaning. In the phenomenon did not appear: The courses on composition, the word re- two stimuli were followed by redundant is sometimes used to refer to sponses of equal magnitude. Pavlov the type of error manifested by a therefore spoke of the more intense CS speaker who proclaims that "We've as ''obscuring" or "overshadowing" won four straight in a row." In a row the less intense CS. conveys the same information as Years later, at the Miami Symposium straight, and the repetition is jarring. on the Prediction of Behavior, Kamin (Sometimes, however, a more remote (1968) presented a series of experi- and less conspicuous redundancy is ef- mental comparisons that provided con- fective in expository writing.) vincing evidence not only for over- Blocking and overshadowing have shadowing but also for another phe- also been observed with operant be- nomenon, also based on the simulta- havior. The primary difference in pro- neous presentation of two stimuli. In cedure is that instead of the stimulus this case, the dominance of one stim- control being based exclusively on the ulus over the other was not established presence versus the absence of a CS, by its intensity but by prior training. If as in respondent conditioning, in op- the subject was trained first with a sin- erant work the discrimination is rec- gle CS and later with a compound that ognized, and both a positive stimulus included the original and some other and an explicit negative stimulus are stimulus, the added CS turned out to normally provided. In a study by vom be ineffective. Kamin spoke of the pri- Saal and Jenkins (1970), for example, or training with the first stimulus as pecking was initially reinforced in the "blocking" the subsequent acquisition presence of green illumination of the of control by the second stimulus. In key but not in the presence of red; later 258 JAMES A. DINSMOOR

a tone and a noise were added to the quently in everyday life and that over procedure, covarying with the first pair a period of time a good many illustra- of stimuli; and finally, control by the tions may be found. For the present, tone and noise alone was compared one example that comes to mind is a with control by those same stimuli in phenomenon, well known to social a group of birds that had not been pre- psychologists, in which individual trained with the red and the green. The members of an outgroup who share a prior training on the red-green discrim- prominent physical characteristic are ination was found to have blocked the perceived as "all looking alike." (For acquisition of control by the tone and a recent list of citations, see Anthony, the noise. Copper, & Mullen, 1992.) That is, for In a study of overshadowing, Miles each individual, the characteristic com- and Jenkins (1973) reinforced pecking mon to all members of the group over- in the presence of a bright light and a shadows the characteristics distinctive tone of 1000 Hz (S+) but not in the of that one individual and makes dif- presence of a light of lower intensity ferent members of the group more dif- and a white noise (S-). The intensity ficult to discriminate. of the second light varied from group Two major types of explanation to group. In subsequent tests, the tone have been offered for blocking and had relatively little influence over the overshadowing. Rescorla and Wagner responding by birds in the group that (1972) have proposed an elegantly had received the largest difference in simple mathematical model of Pavlov- illumination (brightest light vs. total ian conditioning, suggested directly by darkness) but exerted much more con- Kamin's (1968) data, that accounts for trol with birds that had been trained these phenomena in terms of a ceiling with smaller differences in illumina- on the level of associative strength. tion. The easier the light discrimina- When one stimulus is relatively intense tion, the more it overshadowed the or is presented earlier in training, the tone. The harder the light discrimina- overall strength of conditioning based tion, the less it overshadowed the tone. on that stimulus approaches an asymp- Note that in this experiment the vari- tote, and little additional conditioning able that determined the amount of can occur to a second stimulus that is overshadowing was not the absolute less intense or is introduced later in intensity of the S+ (which remained training. Theoretical descriptions of the same from group to group) but the this type (see also Revusky, 1971), size of the difference between that S+ however, do not account for instances and the corresponding S- (stimulus in which overshadowing has been disparity). This may have been the crit- demonstrated on the first trial of con- ical variable in the respondent work as ditioning (e.g., Mackintosh & Reese, well, because the absolute intensity of 1979). the CS (which was assumed to be the The other type of explanation that relevant dimension) was always con- has been popular in discussions of founded with the magnitude of its dif- blocking and overshadowing relates ference from the absence of the CS the reduction in effectiveness to a fail- (disparity). Recognizing that the con- ure to observe or to attend to the sec- ditional stimulus is a discriminative ond stimulus or pair of stimuli (e.g., stimulus enables us to formulate broad- Mackintosh, 1974, pp. 585ff.). Some er, more general principles of behavior. support for this interpretation may be Blocking and overshadowing have found in experiments on observing be- figured prominently in theoretical dis- havior. Two replications of such an ex- cussions about Pavlovian conditioning, periment were conducted in my labo- but at a more applied level they have ratory. In the first one, some of the data largely been ignored. I suspect that were disrupted when we were required their manifestations appear fairly fre- to remove the birds from their usual STIMULUS CONTROL 259 living quarters. Then the second one played on the key, the rate of pecking became to a certain extent redundant dropped more than when the colors scientifically when a similar study was that differed by a small amount were published by Blanchard (1977). In our displayed. experiments, which were somewhat At a broad and general level, then, a more complex than Blanchard's, we model based on principles derived provided the pigeon with an observing from the study of observing can ac- key that altered the tilt of a white line count for the effects of continued train- displayed on a colored ground. During ing, as well as for the overshadowing periods when food was being pro- sometimes noted on the initial trial. On grammed on a variable-interval sched- the other hand, it is difficult to see how ule, pecking the observing key inter- any model based on the observation of mittently produced a shift in the ori- or attention to one stimulus rather than entation of the line, rotating it 450 from another could handle the results ob- the horizontal; during periods when no tained by Revusky (197 1), which mim- food was scheduled (extinction), the ic blocking and overshadowing despite shift was in the opposite direction. the absence of any temporal overlap Meanwhile, the colored backgrounds between the two conditional stimuli. switched back and forth on a schedule Perhaps no single factor can account that was unrelated to the schedules of for all of the data obtained in experi- food delivery and the direction of the ments on blocking and overshadowing. rotation produced by pecking the observing key. On some sessions the al- Progressive Discrimination ternating colors differed by a large amount, and on other sessions they dif- Another phenomenon that highlights fered by a small amount. the importance of the magnitude of the After the rate of pecking had stabi- difference between the stimuli to be lized, however, we synchronized the discriminated (disparity) is known var- alternations of the colors with the al- iously as progressive discrimination ternations between the variable-inter- training, transfer along a continuum, or val schedule and extinction and with the easy-to-hard effect. It has long the direction taken by the line. At this been recognized that the most efficient point, the changes in the orientation of way to train a subject to discriminate the line produced by pecking the ob- between two stimuli that lie close to- serving key became completely redun- gether along the physical continuum is dant: Whenever such a change was to begin with stimuli that are farther produced, its relation to the delivery of apart and then, in successive steps, to food was exactly the same as that of a close the gap (e.g., James, 1890, p. color already displayed on the key. As 515; Montessori, 1912, p. 184). In oth- would be expected, the rate of pecking er words, initial training with easy the observing key in the presence of stimuli from the same dimension ac- one of the S - colors (pecking that pro- complishes more than an equal period duced only S- tilts) dropped precipi- of training with the difficult pair on tously. Much more important was the which the subject is eventually to be fact that the rate of pecking in the pres- tested. In Pavlov's laboratory, this ence of one of the S + colors also de- technique was employed to train a dog clined. In other words, less observing to discriminate between a light gray behavior was maintained by the line and a white circle and to train another tilts as conditioned reinforcers when to discriminate between a fairly well- they covaried with another, more effec- rounded ellipse and a perfect circle tive set of stimuli. Moreover, in accord (Pavlov, 1927/1960, pp. 121ff.). with conventional experiments on Even before bar-pressing or key- overshadowing, when the colors that pecking techniques came into common differed by a large amount were dis- use, systematic data were reported with 260 JAMES A. DINSMOOR

relatively crude procedures, typically instruction. Early papers (Cook, 1960; involving choices between test cards of Skinner, 1958) describe a procedure differing shades of gray. The study that then known as "vanishing": In succes- has been most widely cited, historical- sive steps, letters were eliminated from ly, is one reported by Lawrence (1952). words to be spelled by the subject, In his procedure, the rat was forced to words were deleted from passages to jump across a gap that separated the be recited from memory, or labels for starting box from a pair of goal com- various geographical features were re- partments lined with cardboard in dif- moved from a map. The subject con- fering shades of gray. One group of tinued to respond correctly, despite the subjects was trained throughout with elimination of the original stimuli, two highly similar grays, the same as even when the only remaining cues those used in the final test; two groups might be those generated by chains of started with the lightest and the darkest ongoing behavior. gray and were shifted to the final test Almost immediately, there was a pair at different points in their training; change in terminology, and the same and finally, one group received its first technique came to be known as "fad- 10 trials with the most disparate pair of ing" (Holland, 1960). Then, not long grays, the next 10 trials with a less dis- after that, investigators using nonhu- parate pair, the third 10 trials with a man subjects began studying fading in still less disparate pair, and its last 50 the conditioning laboratory. Using a trials with the final test pair. All the highly specialized training procedure, groups that began with the large dif- Terrace (1963) had initially established ference performed better during their a discrimination between red and green last 50 trials than did the group that illumination of the pigeon's key. Al- began with the difficult discrimination. though he did not point out the paral- The last group, which shifted in a se- lel, the red and the green were equiv- ries of steps, made the fewest errors of alent to the stimuli that were already all the groups. Lawrence suggested that effective when a human trainee began the easy stimuli facilitated later learn- his or her program of instruction. Then ing of the hard stimuli because they a vertical line was superimposed on the helped "the animal to isolate function- red S + and a horizontal line on the ally the relevant stimulus dimension green S -. Gradually, in successive from all the other background and ir- steps, the brightness of the red and the relevant cues" (Lawrence, 1952, p. green was reduced, until the key be- 516). Both the magnitude of the dis- came totally dark except for the lines. parity between the positive and the In other words, the red and the green negative stimulus and the salience of illumination were faded. As the sali- the positive stimulus, which is con- ence of the colors was reduced, the founded with its disparity in many of birds made less and less use of those these studies, contribute to the effec- stimuli and gradually came to depend tive reinforcement of relevant observ- on the direction taken by the line. In ing behavior (Dinsmoor et al., 1983). two replications of this transfer procedure, a total of 4 birds learned the ver- Fading tical-horizontal discrimination without ever pecking in the presence of the S - In contrast to the research on block- (i.e., without making a single error). ing and overshadowing, which was Other techniques, used as experi- concerned exclusively with an eluci- mental controls, proved to be far less dation of the basic principles of con- effective. For example, if the red and ditioning, the original impetus to re- the green were completely removed search on fading seems to have ema- from the key at the time that the ver- nated from attempts to fashion effec- tical and the horizontal lines were in- tive techniques for use in programmed troduced, so that there was no temporal STIMULUS CONTROL 261 overlap between the two sets of stim- Later, when the lines were introduced, uli, hundreds of S- responses occurred the birds had already learned to look at before the birds learned the vertical- and attend to stimuli displayed on the horizontal discrimination. When the key. The lines were displayed in the two sets of stimuli overlapped in time same general location as the colors but nothing was done during this time (i.e., on the surface of the key). Then, to reduce the effectiveness of the red as the salience of the colors was re- and the green, intermediate results duced, their value as reinforcers de- were obtained. Something was evi- clined (Dinsmoor et al., 1982, 1983) dently learned with regard to the lines and the bird shifted from observing the during the period when they were su- colors to observing the lines. perimposed on the colors, but not as Evidence confirming this interpreta- much. It was the gradual reduction in tion comes from a study conducted by the salience of the colors that forced Fields (1978). Again the pigeons were the birds in the fading group to switch pretrained to discriminate between red to the lines during the period when and green illumination of the key. both sets of stimuli were available. By Then white lines of differing orienta- the time the colors were completely tion were superimposed on the red and gone, the birds had already learned to green backgrounds. Between blocks of discriminate accurately on the basis of training trials, Fields inserted probe tri- the lines. (For an alternative procedure als on which he presented the same in which the original stimuli were de- white lines, but on a dark background. layed in onset rather than reduced in These trials enabled him to track the salience, see Touchette, 1971.) acquisition of the line discrimination There is another parallel that was not during the course of training. obvious at the time Terrace conducted The most revealing aspect of Fields' his work: His red and green illumina- (1978) procedure was that with one tion were also comparable to the pre- group of subjects he faded both the red trained or inherently more effective and the green, with another group he stimuli subsequently used in operant faded only the red (S+), leaving the studies of blocking and overshadow- green at its original intensity, and with ing. In light of this correspondence, a third group he faded only the green Terrace's (1963) fading procedure can (S-), leaving the red at its original in- be viewed as a method of overcoming tensity. Interpretation of the difference the overshadowing or undoing the in results between the first two of these blocking of one pair of stimuli by an- groups is complex and need not con- other. By reducing the influence of the cern us here, but the interesting finding originally dominant pair, fading allows occurred with the third group. Data the stimuli that have been blocked or from a large number of experiments in- overshadowed to gain effectiveness. dicate that it is the S+ that reinforces The advantage offered by Terrace's observing (for a review of these data, (1963) fading procedure was that dur- see Dinsmoor, 1983). Accordingly, an ing the prior training, highly effective interpretation in terms of the observing stimuli were available on the key. In paradigm predicts, uniquely I think, contrast to the lines, which were more that it is fading of the S+ that is re- localized, the colors covered the entire quired to transfer control of pecking surface of the key, and anywhere the from the colors to the direction taken pigeon might initially look while peck- by the lines. Fading of the S - should ing that key it was likely to see the have no effect. And that is the result color. Early contact with a stimulus that Fields obtained. "Attenuation of correlated with reinforcement pro- the S - alone ... did not produce stim- duced early acquisition and a high lev- ulus control by the line-tilt dimension" el of performance of the behavior by (Fields, 1978, p. 126). His Figure 1 which the bird observed that stimulus. showed no pecking to either S + or S - 262 JAMES A. DINSMOOR probes when it was the green (S-) il- calling it an A or a B. Four different lumination that was faded. varieties of flower were used, and each of these varied also in the shape of its CONCEPT FORMATION leaves, the angle of the leaves to the stem, and the number of leaves on ei- When, in everyday conversation, ther side. For most of the subjects, the people speak of someone "possessing" angle with the stem was the relevant a certain concept, it is taken for granted feature (i.e., the discriminative stimu- that they are referring to some kind of lus), determining whether A or B was unobservable content or state of the the response to be reinforced. Trabasso mind. But when we examine the mat- found that such strategies as holding ter, we find that the observation that other features constant, rather than al- leads them to use the term concept is lowing them to vary, and, as in ordi- that the person responds in the same nary discrimination training, increasing way to a set of objects or relations that the difference between the positive and have some characteristic or set of char- the negative angles (i.e., disparity), or acteristics in common but does not re- increasing their difference from the spond in that same way to other objects background stimulation by coloring or relations. All birds are called birds, them red (i.e., salience) all led to faster for example, but no snakes, squirrels, acquisition of the correct categoriza- or sheep. As Keller and Schoenfeld put tion. These findings suggest that con- it, "Generalization within classes and cept formation is very similar to dis- discrimination between classes" (1950, crimination training. In particular, the p. 155). effect of highlighting the angles to in- If we can verbalize what it is that crease their salience indicates that a the various instances have in common, major part of the subject's task is to we may be able to define the concept, learn to observe the appropriate feature but often this can be very difficult. Try within the field of stimulation. defining what is meant even by such Traditionally, as I have indicated, familiar categories as a "dog," for ex- psychologists have treated experiments ample, or a "human being." Are you dealing with the formation of discrim- sure you did not include wolves or inations and experiments dealing with coyotes in the first instance or chim- the formation of concepts as two sep- panzees in the second? Did you in- arate and distinct fields of inquiry. But clude feral children? Definition is sec- the close relation between the two has ondary to the behavior: In essence, it long been recognized by some writers is a description of that behavior, al- (e.g., Keller & Schoenfeld, 1950), and though it can provide in turn a rule for whether the traditional distinction is increasing the accuracy of that behav- justified is open to question. ior or for passing it along to another It is true that most of the work on person (e.g., Skinner, 1969, pp. 121- discrimination learning has been car- 125, 136-142). ried out with rats, pigeons, and mon- Laboratory research on the forma- keys, whereas most of the experiments tion of concepts arose from an interest on concept formation have been con- in the processes of human thought and ducted with human subjects. This re- has continued for many years in almost flects the differences in the historical complete isolation from research on origins and rationales of the two types the formation of discriminations. The of research. But there is no hard and general nature of the procedure is il- fast rule: Sometimes.humans are used lustrated by an experiment conducted to study discrimination learning, and by Trabasso (1963). College students sometimes other species are used to were presented with drawings of flow- study the acquisition of concepts. ers, in a randomized sequence, and Another common difference be- were asked to classify each drawing by tween the two types of experiment is STIMULUS CONTROL 263 probably a result of the first. Because ject are deliberately varied. In this re- an entire set of verbal responses can spect, experiments on concept forma- readily and quickly be established with tion are more closely related to what is members of the human species, human colloquially called "real life." They subjects may be asked to acquire sev- provide us with a persuasive bridge be- eral different concepts at the same tween the stripped down simplicity of time. This practice enables the experi- the laboratory and the rich complexity menter to collect more data from each of the natural setting. subject. When other species are used, however, it takes much longer to estab- IMITATION lish each response topography, and the use of any substantial number of cate- There is another form of stimulus gories becomes impractical, even in control that is not well understood but those experiments designated as stud- that merits consideration because it is ies of concept formation. To respond or believed to play an enormous role in not to respond is usually the question. training the young of many species, in- A third feature, which may also re- cluding our own, to behave like their late to the historic use of human sub- parents and other adults. The observa- jects, is that in experiments on concept tion that leads us to speak of imitation formation the distinction between the is that the stimulus setting the occasion positive and the negative stimulus is for a particular response or set of re- often very subtle. Human subjects sponses is the performance of approx- learn simple, gross discriminations so imately the same pattern of behavior rapidly that variations in their perfor- by another individual, usually of the mance created by manipulating some same species. Such a correspondence independent variable might be too between two patterns of behavior is so small to detect, and subtle differences common that it is frequently assumed are necessary to collect meaningful that it must be the result of some in- data. In this case, the historical feature trinsic connection between the behav- may have been extended to serve as an ior of the second organism and that of implicit definition for classifying the the first. But not all behavior is imitat- two types of research: Even when non- ed. In most cases in which imitation is human species are used, subtlety of the said to occur, the correspondence be- properties distinguishing between the tween the two sets of behavior appears positive and the negative instances to be historical in origin. It is a product (e.g., Herrnstein, Loveland, & Cable, of the contingencies of reinforcement. 1976) appears to be characteristic of It has sometimes been suggested those experiments that are considered (e.g., Bandura, 1965) that imitation is to be examples of concept formation. an alternative mode of learning, by But this is a difference in a parametric means of which a response can be es- value, not a difference in the underly- tablished without the use of reinforce- ing process. ment ("no-trial learning"). In some Finally, the most meaningful differ- sense, at a global level, this may be ence between discrimination and con- true, but it is not clear that any new cept formation may be that in experi- principle of behavior is required to ex- ments on discrimination, all aspects of plain Bandura's results. the environment other than the alter- In exploring this issue, it may be nation between the positive and the helpful to distinguish between a com- negative stimulus are kept as uniform plex pattern of physical activity like as possible ("held constant"), so that swimming the breast stroke, riding a they will produce the least possible bicycle, driving an automobile, or per- fluctuation in the results. In experi- forming a new dance step, and the con- ments on concept formation, by constituent movements that make it up trast, other aspects of the stimulus ob- (see Guthrie, 1952, pp. 27-28, or Skin- 264 JAMES A. DINSMOOR ner, 1953, p. 94). Modeling is akin to ginning of the stem, with the follower verbal instruction, and, indeed, might in a second box immediately behind it. be considered another form of the same When the leader was released from its process. If the constituent responses start box, it promptly ran to the choice are already a part of the individual's point at the intersection of the T and repertoire (by prior differentiation) and turned in the designated direction. At some kind of imitative control has al- the end of the arm, it received its usual ready been established, an appropriate allotment of food. The follower rat was sequence or an effective combination released immediately behind it. If the of responses can often be called forth follower rat turned in the same direc- by the corresponding behavior of a tion as its leader, its response was re- model. Sometimes this is all that is inforced with food in a special recep- needed. In other cases, it may still tacle uncovered only after the leader serve as the initial step: Once some ap- rat had passed over it. If the follower proximation to the appropriate behav- rat turned in the other direction, how- ior has occurred, the overall pattern ever, its response was not reinforced. can be maintained and can subsequent- For the follower rat, then, the leader ly be refined by selective reinforce- turning to the right was a discrimina- ment. But if the constituent responses tive stimulus for turning to the right, are not in the individual's repertoire to and the leader turning to the left was a begin with, they cannot be established discriminative stimulus for turning to simply by imitation. Who among us the left. As might be expected, the fol- has not struggled in vain to match the lower rats had no difficulty in master- sounds that have been demonstrated to ing this discrimination. With respect to us by the teacher of a foreign language, this one response, their behavior only to find that the closest approxi- matched or imitated the behavior of mations within our vocal repertoire their leaders. However, in terms of leave much to be desired? Who among learning principles, there was nothing us has not tried to copy someone else's unique about the resulting correspon- drawing, only to discover that much dence between the behavior of the 2 more training was necessary before a animals. To demonstrate this point, satisfactory replica could be produced? Miller and Dollard (1940) trained an- The process by which the correspon- other group of follower rats in a pattern dence between stimulus and response of stimulus control that was exactly the is originally established was illustrated opposite of imitation: Turn left when more than half a century ago in an ex- the leader turns right and turn right periment by Miller and Dollard (1940). when the leader turns left. The two All training was conducted on a special groups learned their tasks with equal maze, shaped like the letter T. First, facility. Following a partner in ball- several "leader" rats were trained to room dancing may provide an illustra- discriminate between black cards and tion of this reversed relationship, but white cards, half of them to enter the such a pattern of stimulus control is not arm of the maze with the black card commonly reinforced in the outside and half to enter the arm with the world. white. The only function of these lead- Apparently there is more to imita- er rats, however, was to provide dis- tion, however, than establishing discriminative stimuli for the rats that criminative control over a single re- were to follow them. sponse. When a number of correspon- When the leader rats had learned to dences have been reinforced between turn in the direction indicated by the the actions of an experimental subject appropriate card, they were used to and the actions of a model, the corre- train the "follower" rats to imitate a spondence itself may become a gov- specific item of behavior. First, a leader erning factor in the relation between was placed in the start box at the be- the two actions, extending to new to- STIMULUS CONTROL 265 pographies of behavior. Baer, Peterson, English words by each of 2 autistic and Sherman (1967) demonstrated this children, in response to those same with 3 profoundly retarded children words presented by an experimenter. who originally showed no tendency to Then some Norwegian words were imitate. In the beginning, the experi- slipped into the sequence, without re- menters used manual guidance and se- inforcement; with continued reinforce- lective reinforcement with food to es- ment of the English words, the pronun- tablish a series of physical actions like ciation of the Norwegian words grad- raising the left arm, tapping a table, or ually improved. These data suggest moving the arm in a circular motion, that the increasing similarity between each of which followed a like action the sounds produced by the child and by the experimenter. the sounds presented by the experi- One of the things these authors menter had itself become reinforcing, showed was that the behavioral corre- leading to successive approximations spondence could itself be placed under of those sounds by the children. the control of some other stimulus in what was therefore known as a condi- STIMULUS EQUIVALENCE tional discrimination. In their study, reinforcement of the child's action was In his presidential address to the conditional upon the prior presentation Midwestern Psychological Associa- of the verbal stimulus, "Do this," fol- tion, Skinner (1950) included brief re- lowed by a demonstration of the de- sumes of several pieces of work con- sired response. Stimuli in the presence ducted in his laboratory that were nev- of which correspondence is reinforced er subsequently reported in greater de- determine when imitation will occur tail. Among these was a procedure that and when it will not occur. In some he called "matching to sample," pre- studies, the choice of a model has de- sented in an attempt to strip some of termined when correspondence would the surplus meaning from what would be reinforced. nowadays be called "cognitive" de- In the Baer et al. (1967) study, 130 scriptions of choice and other complex different responses were employed. patterns of behavior. Eventually the children began to imi- The pigeon was confronted with tate new actions, demonstrated for the three keys, arranged in a horizontal first time. Some of these were repeated row. First, on a given trial, the middle on a number of test trials, still without key was illuminated with a sample col- reinforcement, interspersed among or, perhaps red or green. When the bird training trials on which other actions pecked that key, the sample color was were reinforced. The performance of extinguished and the keys on either these nonreinforced actions continued side were lighted with the comparison to depend, however, on the general ten- colors. One of these colors was the dency to perform responses demon- same as the sample, and the other was strated by the experimenter. When re- different. In a strict matching-to-sam- inforcement of the main body of re- ple procedure, it was a peck on the sponses was replaced by differential same-color key that was reinforced. In reinforcement of other behavior, the what later came to be known as oddity never-reinforced test actions also matching, it was a peck on the differ- dropped out. Both sets of behavior ent-color key that was reinforced. With were restored with restoration of the a delay introduced between the dark- original reinforcement schedule. ening of the center key and the lighting Similarity to the behavior of the of the side keys, Skinner's procedure model may also serve as a conditioned was widely used by more cognitively reinforcer. In another experiment, Lo- oriented psychologists to analyze the vaas, Berberich, Perloff, and Schaeffer processes involved in responding to (1966) reinforced the production of past events (memory). 266 JAMES A. DINSMOOR

Years later, Sidman (1971, 1994) the alternative stimuli (in this case, red adapted Skinner's procedure to the task or green comparison stimuli) in a dis- of teaching a severely retarded boy to crimination is positive (i.e., followed read English text. Prior to the proce- by reinforcement of the response). The dure in question, this boy had already relation between comparison stimulus learned correspondences in both direc- and reinforcer is selected by or condi- tions between spoken words and pic- tional upon the nature of the sample tures. That is, he had learned to choose stimulus. When the sample stimulus is the pictures that illustrated the words red, pecking red is reinforced and presented as sample stimuli and to pro- pecking green is not reinforced; when duce the correct vocal responses to the sample is green, pecking red is not (name) the pictures. During the train- reinforced and pecking green is rein- ing procedure, the sample stimuli were forced. words spoken by the experimenter, and Sidman (1994) has also suggested the comparison stimuli were printed the use of the term conditional stimu- words. Somewhat to Sidman's surprise, lus to refer to the role of the sample after his subject learned to choose the stimulus in this type of discrimination, printed word that corresponded to each but such a usage could lead to termi- spoken word, not only was he able to nological confusion, because that term proceed in the opposite direction, pro- is widely used in Pavlovian condition- ducing the correct oral responses to ing to refer to the stimulus that ac- printed text (reading aloud) but, with- quires its effectiveness through its tem- out further training, was able to choose poral relation to the unconditional pictures corresponding to the printed stimulus. The terms instructional stim- words (reading comprehension). Estab- ulus (Cumming & Berryman, 1965) or lishing an equivalence between the au- contextual stimulus (Sidman, 1994) ditory and the visual forms of a series have also been suggested; the former of words had extended the subject's term has intuitive appeal, and the latter ability to select the correct pictures term has the advantage of emphasizing from the original spoken words to their that stimuli and responses usually do printed equivalents. not pair off in a simplistic one-to-one Struck by these initial findings, Sid- correspondence, as sometimes implied man launched a major program of rein early writings by Watson (1919), for search designed to explore their theo- example, but are characteristically re- retical ramifications (Sidman, 1994). lated in a way that depends on what Because there is now a burgeoning lit- other stimuli are present (context). erature on the topic, it may be impor- Although an exposition of the nature tant to consider some terminological of the procedure we are considering issues. First, Sidman objected to the has been greatly simplified by the use use of the term matching to sample as of physically identical colors as ex- a description of Skinner's procedure, emplars, in establishing equivalence on the grounds that this phrase implied classes the experimenter utilizes stim- a generalized behavioral outcome that uli that bear no necessary resemblance did not necessarily result from that pro- to each other and establishes their cor- cedure. In its place, he substituted the respondence through the program of phrase conditional discrimination, first training. A quantity of objects, for ex- applied by Cumming and Berryman ample, and the corresponding numeral (1965). Generically, the term condi- begin as arbitrary, unrelated stimuli, tional discrimination covers more than but the standard relations characteristic the matching-to-sample paradigm (see of a given language can be established Yarczower, 1971), but in the Cumming through this type of training. Eight ob- and Berryman usage an additional, ex- jects can be made equivalent to the traneous stimulus (e.g., a red sample or digit "8" and the binary number a green sample) determines which of "1000" and the sound "ate" and the STIMULUS CONTROL 267 printed word "eight" and even the likelihood of class formation, the for- Spanish word "ocho," thus forming a mal characteristics of specific emergent class, all of whose members may be relations, and the degree of transfer be- said for our purposes to be equivalent. tween stimuli. Nodes are formally de- When they become members of the fined as individual stimuli that are same class, they are for many purposes linked by the program of training to interchangeable. If we respond in the more than one other stimulus within same way to these physically very dif- the particular equivalence class; they ferent stimuli, they may be said to have may be thought of as intervening steps the same meaning. or mediators of the relation between Sidman (1994) has laid out three cri- stimuli that were not presented togeth- teria, all of which are necessary to the er during the training. And nodal dis- mathematical definition of an equiva- tance refers to the number of such lence relation among a set of stimuli. steps that lie between the stimuli that Reflexivity requires that, without spe- are to be tested. In a number of studies, cific instruction or training, the subject the authors found, this parameter had a choose each stimulus in the list as a consistent influence on the subjects' comparison in reaction to that same performances on a number of different stimulus as a sample (i.e., generalized types of test. A point of emphasis in identity matching). Symmetry requires their review was that functions ac- that the subject perform correctly when quired by one stimulus in an equiva- the roles of sample and comparison lence class do not transfer equally, but stimuli are reversed. Transitivity in- rather volves three stimuli in a sequence. differentially, to other members Once "if a, then b" and "if b, then c" of the class. The relatedness of the have been established, then "if a, then stimuli is affected by the directionality c" must emerge without further in- of training and is an inverse function struction or training. of nodal distance. The apparent significance of equiv- The relation between the formation alence classes for what are sometimes of equivalence classes and earlier, sim- known as higher mental processes is pler forms of discrimination learning currently attracting a considerable remains the subject of wide-ranging amount of attention within the behav- discussion. Still in dispute, for exam- ior-analytic community. Quite a bit of ple, is the question of whether emer- research has recently been reported, ac- gent relations and stimulus classes can companied by a large amount of theo- be demonstrated with nonhuman sub- rizing. However, it remains a relatively jects (see Schusterman & Kastak, new field of investigation, still in flux, 1993; Zentall & Urcuioli, 1993). Sid- and the available information does not man (1994) has suggested that the for- as yet permit us to be sure how it fits mation of equivalence classes through into a broader, more systematic frame- conditional discrimination training work. may not be reducible to or explicable One active program of research by other processes with which we are stems from Fields and Verhave's already familiar but may be a wholly (1987) analysis of the parameters that independent phenomenon grounded in define the internal structure of an the evolution of the human species. For equivalence class. In a review of the those of us in search of systematic laws relevant literature, Fields, Adams, and of behavior, this is not a very satisfying Verhave (1993) examined the effects of solution. Hayes (1991), on the other two of these parameters. Directionality hand, has suggested that equivalence refers to whether a stimulus serves as classes can be traced to a preexperi- a sample or a comparison in the pro- mental history of training in relational cess of training. In several studies, this responding. At the present time, such relation was found to influence the issues are far from being resolved. 268 JAMES A. DINSMOOR

REFERENCES N. Chase (Eds.), Dialogues on verbal behavior (pp. 19-40). Reno, NV: Context Press. Anthony, T., Copper, C., & Mullen, B. (1992). Herrnstein, R. J., Loveland, D. H., & Cable, C. Cross-racial facial identification: A social cog- (1976). Natural concepts in pigeons. Journal nitive integration. Personality and Social Psy- ofExperimental Psychology: Animal Behavior chology Bulletin, 18, 296-301. Processes, 2, 285-302. Baer, D. M., Peterson, R. E, & Sherman, J. A. Holland, J. G. (1960). Teaching machines: An (1967). The development of imitation by re- application of principles from the laboratory. inforcing behavioral similarity to a model. Journal of the Experimental Analysis of Be- Journal of the Experimental Analysis of Be- havior, 3, 275-287. havior, 10, 405-416. Honig, W. K., Thomas, D. R., & Guttman, N. Bandura, A. (1965). Vicarious processes: A (1959). Differential effects of continuous ex- case of no-trial learning. In L. Berkowitz tinction and discrimination training on the (Ed.), Advances in experimental social psy- generalization gradient. Journal of Experi- chology (Vol. 2, pp. 1-55). New York: Aca- mental Psychology, 58, 145-152. demic Press. James, W. (1890). Principles of psychology. Blanchard, R. (1977). Control of observing by New York: Holt, Rinehardt, & Winston. antecedent stimuli. Learning and Motivation, Johnson, D. F (1970). Determiners of selective 8, 569-580. stimulus control in the pigeon. Journal of Blough, D. S. (1969). Attention shifts in a Comparative and Physiological Psychology, maintained discrimination. Science, 166, 125- 70, 298-307. 126. Kamin, L. J. (1968). "Attention-like" processes Cook, D. A. (1960). On vanishing stimuli in in classical conditioning. In M. R. Jones (Ed.), instructional material. Journal of the Experi- Miami symposium on the prediction of behav- mental Analysis of Behavior, 3, 292. ior, 1967: Aversive stimulation (pp. 9-3 1). Cumming, W. W., & Berryman, R. (1965). The Coral Gables, FL: University of Miami Press. complex discriminated operant: Studies of Keller, F S., & Schoenfeld, W. N. (1950). Prin- matching-to-sample and related problems. In ciples ofpsychology: A systematic text in the D. I. Mostofsky (Ed.), Stimulus generalization science of behavior. New York: Appleton- (pp. 284-330). Stanford, CA: Stanford Uni- Century-Crofts. versity Press. Lawrence, D. H. (1952). The transfer of a dis- Dinsmoor, J. A. (1983). Observing and condi- crimination along a continuum. Journal of tioned reinforcement. Behavioral and Brain Comparative and Physiological Psychology, Sciences, 6, 693-704. 45, 511-516. Dinsmoor, J. A., Bowe, C. A., Dout, D. L., Mar- Lovaas, 0. I., Berberich, J. P., Perloff, B. F, & tin, L. T., Mueller, K. L., & Workman, J. D. Schaeffer, B. (1966). Acquisition of imitative (1983). Separating the effects of salience and speech by schizophrenic children. Science, disparity on the rate of observing. Journal of 151, 705-707. the Experimental Analysis of Behavior, 40, Mackintosh, N. J. (1974). The psychology of 253-264. animal learning. London: Academic Press. Dinsmoor, J. A., Mueller, K. L., Martin, L. T., Mackintosh, N. J., & Reese, B. (1979). One- & Bowe, C. A. (1982). The acquisition of trial overshadowing. Quarterly Journal ofEx- observing. Journal of the Experimental Anal- perimental Psychology, 31, 519-526. ysis of Behavior, 38, 249-263. Miles, C. G., & Jenkins, H. M. (1973). Over- Dinsmoor, J. A., Sears, G. W., & Dout, D. L. shadowing in operant conditioning as a func- (1976). Observing as a function of stimulus tion of discriminability. Learning and Moti- difference. Journal of Experimental Psychol- vation, 4, 11-27. ogy: Animal Behavior Processes, 2, 154-162. Miller, N. E., & Dollard, J. (1940). Social Fields, L. (1978). Fading and errorless transfer learning and imitation. New Haven, CT: Yale in successive discriminations. Journal of the University Press. Experimental Analysis of Behavior, 30, 123- Montessori, M. (1912). The Montessori method. 128. New York: Frederick A. Stokes. Fields, L., Adams, B. J., & Verhave, T. (1993). Pavlov, I. P (1960). Conditioned reflexes: An The effects of equivalence class structure on investigation of the physiological activity of test performances. Psychological Record, 43, the cerebral cortex (G. V. Anrep, Trans.). New 697-712. York: Dover. (Original work published 1927) Fields, L., & Verhave, T. (1987). The structure Pierrel, R., Sherman, J. G., Blue, S., & Hegge, of equivalence classes. Journal of the Exper- F W. (1970). Auditory discrimination: A imental Analysis of Behavior, 48, 317-332. three-variable analysis of intensity effects. Guthrie, E. R. (1952). The psychology of learn- Journal of the Experimental Analysis of Be- ing (Rev. ed.). New York: Harper & Row. havior, 13, 17-35. Hanson, H. M. (1959). Effects of discrimination Rescorla, A. A., & Wagner, A. R. (1972). A training on stimulus generalization. Journal of theory of Pavlovian conditioning: Variations Experimental Psychology, 58, 321-334. in the effectiveness of reinforcement and non- Hayes, S. C. (1991). A relational control tieory reinforcement. In A. H. Black & W. F Prokasy of stimulus equivalence. In L. J. Hayes & P (Eds.), Classical conditioning: Vol. 2. Current STIMULUS CONTROL 269

research and theory (pp. 64-99). New York: Touchette, P E. (1971). Transfer of stimulus Appleton-Century-Crofts. control: Measuring the moment of transfer. Revusky, S. H. (1971). The role of interference Journal of the Experimental Analysis of Be- in association over a delay. In W. K. Honig & havior, 15, 347-354. P. H. R. James (Eds.), Animal memory (pp. Trabasso, T. (1963). Stimulus emphasis and all- 155-213). New York: Academic Press. or-none learning in concept identification. Schusterman, R. J., & Kastak, D. (1993). A Journal ofExperimental Psychology, 65, 398- California sea lion (Zalophus californianus) is 406. capable of forming equivalence relations. Psy- vom Saal, W., & Jenkins, H. M. (1970). Block- chological Record, 43, 823-839. ing the development of stimulus control. Sidman, M. (1971). Reading and auditory-vi- Learning and Motivation, 1, 52-64. sual equivalences. Journal of Speech and Wagner, A. R., Logan, E A., Haberlandt, K., & Hearing Research, 14, 5-13. Price, T (1968). Stimulus selection in animal Sidman, M. (1994). Equivalence relations and discrimination learning. Journal of Experi- behavior: A research story. Boston: Authors mental Psychology, 76, 171-180. Cooperative. Watson, J. B. (1919). Psychology from the Skinner, B. F (1950). Are theories of learning necessary? Psychological Review, 57, 193- standpoint ofa behaviorist. Philadelphia: Lip- 216. pincott. Skinner, B. F (1953). Science and human be- Yarczower, M. (1971). Stimulus control during havior. New York: Macmillan. conditional discrimination. Journal of the Ex- Skinner, B. F (1958). Teaching machines. Sci- perimental Analysis of Behavior, 16, 89-94. ence, 128, 969-977. Yarczower, M., & Switalski, R. (1969). Stimu- Skinner, B. F (1969). Contingencies of rein- lus control in the goldfish after massed ex- forcement: A theoretical analysis. New York: tinction. Journal of the Experimental Analysis Appleton-Century-Crofts. of Behavior, 12, 565-570. Terrace, H. S. (1963). Errorless transfer of a Zentall, T. R., & Urcuioli, P. J. (1993). Emer- discrimination across two continua. Journal of gent relations in the formation of stimulus the Experimental Analysis of Behavior, 6, classes by pigeons. Psychological Record, 43, 223-232. 795-810.