CONTROL BY CONTEXTUAL STIMULI IN NOVEL SECOND-ORDER CONDITIONAL DISCRIMINATIONS

：芭芭拉

CONTROL BY CONTEXTUAL STIMULI IN NOVEL SECOND-ORDER CONDITIONAL DISCRIMINATIONS
Luis Antonio Pérez-González, Héctor Martínez. The Psychological Record. Gambier: Winter 2007. Vol. 57, Iss. 1; pg. 117, 27 pgs

Abstract (Summary)
The present research shows that, for the contextual stimulus to acquire a particular function, a correspondence between the stimulus and the contingencies must exist, in the sense of determining the sample-comparison relations that will be reinforced (as demonstrated in Study 1). Second, that person must learn the correspondence between the instructions ("Select the name of the person whose discipline is identical to that of the person named in the sample") and the following consequences.

：芭芭拉

Full Text (10524  words)

Copyright The Psychological Record Winter 2007

[Headnote]
Eighteen undergraduates participated in studies designed to examine the factors that produce transfer of contextual functions to novel stimuli in second-order conditional discriminations. In Study 1, participants selected comparison B1 given sample A1 and comparison B2 given sample A2 in a matching-to-sample procedure. Contextual stimuli X1 or X2 were introduced next. In the presence of X1, selections of B1 given A1 and selections of B2 given A2 were reinforced; in the presence of X2, selections of B2 given A1 and selections of B1 given A2 were reinforced (the XAB relations). The participants then learned novel conditional discriminations with stimuli E and F. When we presented the contextual stimuli X1 and X2 with EF, all 3 participants selected the same and opposite comparisons, respectively, as those previously established in EF training. This performance demonstrated the transfer of functions of contextual stimuli to a conditional discrimination with novel stimuli. In Study 2, participants were exposed to the same teaching and testing conditions as in Study 1, except that stimuli X1 and X2 were introduced in a single phase. Of the 4 participants, 2 showed transfer to a novel conditional discrimination. In Study 3 the selections in the XAB task had no differential consequences. Study 4 replicated the conditions of Study 1, with the exception that the AB phase was not taught. Finally, in Study 5 the EF phase was not taught. All participants failed to demonstrate contextual control of the novel conditional discrimination. These results demonstrated that a complete sequence of conditional-discrimination teaching plays an important role in the acquisition and transfer of contextual control to novel conditional discriminations. These results are relevant to our understanding of instructional control.


Humans typically learn conditional discriminations as components of many verbal skills. In a conditional discrimination procedure, a sample stimulus and two or more comparison stimuli appear on each trial. The selection of one comparison produces the reinforcer. On other trials, another sample stimulus appears with the comparisons; the selection of the other comparison produces the reinforcer. Figure 1 shows an example of a conditional discrimination with Cervantes and Renoir as samples and Tolstoi and Siqueiros as comparisons. Here, Cervantes goes with Tolstoi, because they share the same discipline (being writers), and Renoir goes with Siqueiros, because they share the same discipline (being painters). Thus, these selections produce the reinforcer. In many occasions in daily life, however, selections among comparisons given a sample depend upon contextual stimuli. Contextual stimulus X1 controls specific sample-comparison relations; but contextual stimulus X2 controls relations between each sample and the alternative comparison. In the example of Figure 1, Cervantes goes with Tolstoi and Siqueiros goes with Renoir only if the contextual stimulus is "Same Discipline"; this happens because Cervantes and Tosltoi, on the one hand, and Siqueiros and Renoir, on the other hand, share the same discipline. If, alternatively, the contextual stimulus is "Different Discipline," then Cervantes goes with Renoir and Siqueiros goes with Tolstoi. Thus, contextual stimuli control sample-comparison selections. For this reason, we speak of contextual control or second-order conditional discriminations (e.g., Sidman, 1986). Many studies have examined the acquisition of second-order conditional discriminations (e.g., Bush, Sidman & de Rose, 1-9-8-9; Dymond & Barnes, 1995; Gatch & Osborne, 1-9-8-9; Hayes, Kohlenberg & Hayes, 1991; Kennedy & Laitinen, 1988; Lynch & Green, 1991; Markham & Dougher, 1993; Meehan & Fields, 1995; Pérez-González, 1991; Pérez-González & Serna, 1993a, 1993b, 2003; Pérez-González, Spradlin, & Saunders, 2000; Roche & Barnes, 1996, 1997; Serna, 1991; Serna & PérezGonzález, 1994, 1997, 2003; Wulfert & Hayes, 1988).

Contextual stimuli should exert functions at higher level than the functions exerted by the samples. Therefore, a procedure that serves to teach second-order conditional discriminations does not guarantee that the putative contextual stimuli be true contextual stimuli. Actually, in some studies on second-order conditional discriminations, the two stimuli presented in a trial may function as a compound to control responding, just as if they were one single stimulus. For example, X1 and A1 may function as a compound controlling selections of comparison B1, in the same way that a single stimulus (i.e., H1 ) could have controlled selections of B1. Thus, stimuli X1 and A1 would have the same level of functions. Therefore, stimulus X1 does not exert the function of a contextual stimulus. The same occurs with stimulus X2. Hence, it is not correct to consider that stimuli X1 and X2 are contextual stimuli (e.g., Lynch & Green, 1991; Pérez-González & Serna, 2003).

The procedure to teach second-order conditional discriminations may also result in stimuli X1 and X2 controlling sample-comparison relations. In such a case, stimuli X1 and X2 would have functions different from the functions of the samples. For testing this possibility, it is necessary to show that putative contextual stimuli control relations with stimuli other than the stimuli taught with them, which is considered as evidence of "true" contextual control (Lynch & Green, 1991). Lynch and Green reported performances of participants that showed that untaught conditional relations emerged under the control of two contextual stimuli that have been paired with the stimuli used as contextual in teaching. For that reason, they claimed that they showed contextual control. PérezGonzález and Serna demonstrated with children an unequivocal case of contextual control by teaching a conditional discrimination with contextual stimuli and testing the two contextual stimuli with novel stimuli of other conditional discrimination (Pérez-González & Serna, 1993a, 1993b, 2003; Serna & Pérez-González, 1994, 1997, 2003; see details below).

The functions of contextual stimuli may transfer to novel stimuli. PérezGonzález and Serna (1993a, 2003) documented that stimuli that acquired the functions of contextual stimuli transferred that function when presented in novel conditional discriminations. Pérez-González and Serna first taught teenagers and adults a sim-ple conditional discrimination AB (i.e., select comparison B1 in the presence of sample A1 and select comparison B2 in the presence of sample A2). Second, these experimenters brought the AB conditional discrimination under the control of contextual stimuli X1 and X2. Thus, in the presence of X1, they reinforced selections of the same comparisons as in conditional discrimination AB (i.e., selecting B1 in the presence of X1 and A1 and selecting B2 in the presence of X1 and A2). In the presence of X2, however, they reinforced the opposite comparisons (i.e., selecting B2 in the presence of X2 and A1 and selecting B1 in the presence of X2 and A2). This constituted the second-order conditional discrimination, XAB. The teaching of the AB and XAB conditional discriminations was the experimental prerequisite for probing the functions of stimuli X1 and X2 as contextual stimuli in the presence of novel stimuli. To prepare for the probe, Pérez-González and Serna first taught another conditional discrimination EF (by reinforcing selections of F1 in the presence of E1 and selections of F2 in the presence of E2). They finally probed the functions of X1 and X2 as contextual stimuli by presenting them in the EF conditional discrimination (without differential reinforcement). The participants selected the same comparison as in the EF task in the presence of contextual stimulus X1 (i.e., they chose F1 in the presence of X1 and E1, and they chose F2 in the presence of X1 and E2). The participants also selected the opposite comparison in the presence of contextual stimulus X2 (i.e., they chose F2 in the presence of X2 and E1, and they chose F1 in the presence of X2 and E2). Thus, Pérez-González and Serna demonstrated that the functions of contextual stimuli transferred from the XAB second-order conditional discrimination to a novel secondorder conditional discrimination involving different stimuli (XEF). In summary, with this study and two additional ones, Pérez-González and Serna demonstrated that the functions of contextual stimuli may transfer to a potentially infinite number of novel conditional discriminations.

Several studies have replicated the results of Pérez-González and Serna (1993a, 2003). For example, Meehan and Fields (1995) used a version of the Pérez-González and Serna's procedure, in which responding with a stimulus analogous to contextual stimulus X2 was taught (but X1 was not taught). The procedure served to produce contextual control in 3 out of 5 adults. Moreover, Carpentier, Smeets, and Barnes-Holmes (2002, 2003) replicated these results with 24 adults using basically the same procedure as Pérez-González and Serna along with other discriminations and more complex designs (i.e., they taught DE, XDE, and FG and they probed XFG). Also, Serna and Pérez-González (2003, Study 1) demonstrated the transfer of contextual functions after allowing adult participants to make arbitrary selections between the comparisons in a conditional discrimination. These experimenters first taught the AB and XAB conditional discriminations. Then they presented the CD conditional discrimination without differential reinforcement, until the participants' selections were stable for a number of consecutive trials. Some participants selected D1 in the presence of C1, for example, and selected D2 in the presence of C2; other participants selected D2 in the presence of C1 and D1 in the presence of C2. Finally, Serna and PérezGonzález showed that the functions of the contextual stimuli X1 and X2 transferred to the CD conditional discrimination.

The processes studied by Pérez-González and Serna (1993a, 2003) showed transfer of specific functions of contextual stimuli. Serna and Pérez-González (2003) demonstrated a more generalized transfer. After showing transfer of the specific functions of the contextual stimuli, they presented novel stimuli Z1 and Z2 in place of the contextual together with conditional discriminations. The novel stimuli controlled conditional discriminations, as if they had acquired some functions of the contextual stimuli. Nonetheless, no specific novel stimulus could control selections of the same comparison as in its absence, because stimuli Z1 and Z2 had not specific relations to X1 and X2. (The same happens with selections of the opposite comparisons as in its absence.) Therefore, 1 participant selected the same comparisons in the presence of Z1 as in its absence and the other participant selected the opposite comparisons. Thus, this study showed that the functions of contextual stimuli transferred in a generalized way. The authors designated this process generalized contextual control.

The studies of Carpentier et al. (2002), and Pérez-González and Serna (1993a, 2003), constituted demonstrations of a basic transfer process when contextual stimuli are presented with novel conditional discriminations. These studies, however, did not analyze the conditions under which such a process occurs. In contrast, the study by Carpentier et al. (2003) included one condition in which the second-order conditional discrimination was taught before the first-order conditional discrimination. Although all 6 participants showed transfer of contextual stimulus functions in the original condition (i.e., when the participants learned the first-order conditional discriminations first), only 4 of 6 participants showed transfer of contextual stimulus functions when they learned the second-order conditional discrimination first. Thus, they demonstrated that teaching the first-order conditional discrimination first facilitates the transfer of specific functions. Serna and Pérez-González (2003) also analyzed the effect of teaching the second first-order conditional discriminations for generalized contextual control to occur. They found generalized transfer of contextual functions only when the first-order conditional discriminations were taught.

The studies by Carpentier et al. (2003) and by Serna and PérezGonzález (2003) showed that the functions of contextual stimuli may not be acquired or transferred under certain conditions. The main goal of the present research was to further study the conditions necessary for acquiring contextual control of specific functions and for transferring this control to novel conditional discriminations. We also analyzed the conditions that promote generalized contextual control. For these purposes, we examined transfer in five studies that used variations of the basic learning sequence of Pérez-González and Serna (2003). Our Study 1 was a replication of Study 1 of Pérez-González and Serna. This replication served as a reference to compare the results of those of the remaining studies. Study 2 analyzed whether the specific procedure for teaching the XAB conditional discrimination affects acquisition and transfer of contextual functions. Study 3 analyzed whether providing differential consequences during XAB teaching was necessary for acquisition and transfer. Study 4 analyzed whether teaching AB was necessary for the acquisition and transfer of contextual functions. Study 5 analyzed whether teaching EF was necessary for transfer of contextual functions.

：芭芭拉

General Method

Participants

Through personal contacts at the School of Psychology of the University of Oviedo, 14 female and 4 male undergraduates ranging from 18 to 37 years of age were recruited. All participants were Spanish speakers. They received 6 euros for their participation. None of the participants had previously experienced conditional discrimination tasks. The participants were randomly assigned to each study.

Apparatus and Setting

The studies were conducted in experimental cubicles of the laboratory of the School of Psychology. Each cubicle contained a personal computer with a Dos-based O p e r ating system that controlled stimulus presentation and data collection. Each participant was seated in front of a computer screen (22 × 20 cm) and keyboard. The stimuli were arbitrary visual forms (approximately 1.5 × 2 cm) identical to those employed by Pérez-González and Serna (2003).

Procedure

Participants attended the laboratory for one continuous session, which lasted less than 25 minutes. Because the experiment was programmed in blocks of trials, each participant passed through several teaching phases and one test.

Types of Conditional Discriminations

We used first-order and second-order conditional discriminations. In the first-order conditional discriminations, two figures served as samples (e. g., A1 or A2) and another two served as comparisons (e.g., B1 and B2). In the presence of A1, B1 was the correct comparison, and in the presence of A2, B2 was the correct comparison. In the second-order conditional discriminations, two figures (X1 and X2) were added as contextual stimuli to the conditional discrimination. The presence of X1 or X2 determined the selection of which comparison would be considered correct in the presence of each sample. In the presence of X1, the correct comparison was the same as in absence of contextual stimuli. In the presence of X2, the correct comparison was the comparison that has been incorrect in the presence of X1. Thus, X1 determined A1-B1 and A2-B2 as correct relations whereas X2 determined A1-B2 and A2-B1 as correct relations.

Presentation of Stimuli

The sample always appeared at the center of the screen, as Figure 2 shows. In trials of the second-order conditional discrimination, the stimulus X1 or stimulus X2 appeared above the sample. The two comparisons appeared in a row below the sample. On each trial all stimuli were displayed simultaneously and no observing response was required. The position of the comparisons varied randomly along trials. The combinations of contextual stimuli and samples were randomly presented in each block of two or four consecutive trials. For example, the X1-A1 and X1-A2 combinations were randomly presented every pair of trials in some phases; also, combinations X1-A1, X1-A2, X2-A1, and X2-A2 were randomly presented on four consecutive trials in other phases (see below).

Stimulus Selection and Consequences

For stimulus selection, a cursor appeared at the bottom left corner of the screen. Pressing the "B" or "N" keys moved the cursor below the left or right comparison respectively. Then, pressing the "H" key moved the cursor upward toward the comparison, delivered the programmed consequence, and ended the trial. During the trials with programmed differential consequences, a correct response produced a sequence of four musical notes whereas an incorrect selection produced a low tone. The participants' responding showed that the sequence of musical notes worked as a reinforcer and the low tone as a punisher. The trials without programmed differential consequences ended immediately after the comparison selection. After each trial ended, the screen went blank for a 2-s intertrial interval.

Instructions

At the start of the session, the following instructions (in Spanish) appeared on the screen:

Thank you for participating in this research. Some figures are going to appear on the screen. Press the B and N keys to move the ^ symbol under the figures. To actually select the figure, use the H key. Hearing some music after your response means that you are right. Hearing a low tone means an error or incorrect response. Sometimes you won't hear anything after your response. Always try to be correct. (Press the space bar when you wish to start).

Questions were answered by repeating the relevant portions of the instructions. The experimenter remained in the room until the participant completed the first three trials; then, the experimenter left the room.

Delayed Prompt Procedure

To teach conditional discriminations, a delayed-prompt procedure (e.g., McIlvane & Dube, 1992; Touchette, 1971) was used in which at first the incorrect comparison stimulus was presented for 1 s and then vanished. The interval during which the incorrect comparison was present was multiplied by 1.3 on any trial following a correct selection, and divided by 1.3 on any trial following an incorrect selection (unless the lower bound of 1 s was reached). Correct selections made after the incorrect comparison had already disappeared increased the interval during which the incorrect comparison was present on the next trial. Only responses made with both comparisons present counted in relation to the criterion for advancing to the next phase of the study.

Conditional-Discrimination Teaching and Testing

Successive series of trials are described below as phases. Each phase contained all possible combinations of samples corresponding to a conditional discrimination. There were teaching and testing phases. Teaching phases were designed to teach or review conditional discriminations. In the teaching phases, a series of 24 trials was repeated until the participant responded correctly on either 8 or 12 consecutive trials (this number depended on the phase; see Table 1). Once the participant reached the relevant criterion, the next phase started. For each teaching phase, the proportion of selections with differential consequences was 1, .5, or .25 (see Table 1). Testing phases were designed to test the emergence of a novel conditional discrimination. In these phases, 24 trials were presented without differential consequences. The session ended after the final programmed phase.

Phases

We designed one or more phases to teach or to test the conditional discriminations. Table 1 summarizes the specific procedures used in each phase of Studies 1 to 5. Conditional discrimination AB was taught in Phases 1 to 3, XAB was taught in Phases 4 to 8, and EF was taught in Phases 9 to 11. In Phase 12, the XEF conditional discrimination probed the transfer of functions of contextual stimuli. Studies 2 to 5 varied in that some of the phases were omitted or modified.

Phases 1, 2, and 3: AB teaching. During Phase 1 (see Table 1), stimuli A1 and A2 were presented randomly as samples across trials; stimuli B1 and B2 served as comparisons. Choosing B1 in the presence of A1 and B2 in the presence of A2 was reinforced. As shown in Table 1, the delayed prompt procedure was used and the participants received the consequences (described above) on each trial. This procedure continued until the participant obtained eight consecutive correct responses with both comparison stimuli present. The program then advanced to the next AB teaching phase. Phases 2 and 3 presented the same discriminations as in Phase 1; the criterion to advance to the next phase was also the same as in Phase 1. The delayed prompt procedure, however, was omitted, and the probability of reinforcement per trial decreased to .5 (Phase 2) and to .25 (Phase 3). Thus, Phases 2 and 3 were conducted to gradually reduce the density of differential consequences across trials in preparation for the unreinforced trials of the test phase.

Phase 4: X1AB teaching. During this phase, contextual stimulus X1 and sample A1 or A2 were presented on each trial. The procedure was identical to that of Phase 1, except that 12 consecutive correct responses were required to advance to the next phase and all correct responses were reinforced.

Phase 5: X2AB teaching. This phase was identical to X1AB teaching (Phase 4), except that the contextual stimulus was X2 instead of X1. Also, selecting B2 in the presence of A1 and selecting B1 in the presence of A2 was reinforced.

Phases 6, 7, and 8: XAB review. During the first phase of XAB teaching (Phase 6) the four combinations of stimuli of Figure 2 were presented randomly across trials. As in the X1AB and X2AB teaching phases, selecting B1 in the presence of A1 and selecting B2 in the presence of A2 was reinforced on the trials that involved X1. Selecting B2 in the presence of A1 and selecting B1 in the presence of A2 was reinforced in the trials that involved X2. After eight consecutive correct selections, participants advanced to the next phase. The next two phases were identical to Phase 6, except that the probability of reinforcement decreased to .5 (Phase 7) and then to .25 (Phase 8).

Phases 9, 10, and 11: EF teaching. EF conditional-discrimination teaching was the same as for the AB conditional discrimination, except that stimuli E1 and E2 were presented as samples and stimuli F1 and F2 were presented as comparisons. Selections of F1 given E1 and selections of F2 given E2 were considered correct. The delayed prompt procedure was introduced again on Phase 9 and removed on Phases 10 and 11.

Phase 12: XEF test. The XEF test consisted of presenting randomly the four XE stimulus combinations (X1-E1, X1-E2, X2-E1, or X2-E2) with stimuli F1 and F2 as comparisons (see Figure 2, lower panel). No differential consequences occurred and the session ended after 24 trials. Selecting F1 in the presence of X1 and E1, or X2 and E2, and selecting F2 in the presence of X1 and E2, or X2 and E1, was defined as correct.

：芭芭拉

Study 1

Study 1 replicated the first study of Pérez-González and Serna (2003) with minor modifications in the procedures. The results of this replication will be compared to those of Studies 2 to 5.

Method

Three participants received the procedure described in the General Method. All teaching phases were used, as shown in Table 1.

Results

Participant Lara required 137 trials (129 correct), Maxo required 153 trials (146 correct), and Mara required 164 trials (153 correct) to complete the entire sequence of phases. They received, thus, few trials above the minimum (the entire teaching and testing sequence required a minimum of 120 trials to be completed, if a participant responded correctly on all trials). They made from 7 to 11 errors. The 3 participants made an error in the first trial with stimulus X2 (Phase 5). They learned the phases with XAB (Phases 4 to 8) within 52 to 60 trials.

Individual selections on each of the 24 test trials of the XEF test are shown in Table 2. Participants Lara and Maxo responded correctly on 24 trials of the XEF test and Mara responded correctly on 23. Thus, all 3 participants transferred contextual control to a new conditional discrimination.

Discussion

These results replicated earlier findings that demonstrated transfer of specific contextual-control functions by showing that the functions of the contextual stimuli transferred to novel conditional discriminations (Carpentier et al., 2002, 2003; Pérez-González & Serna, 2003). All participants in the experimental conditions of these three studies and of the present one demonstrated transfer of contextual functions. Moreover, other participants in another study (Serna & Pérez-González, 2003) showed similar results. Because the studies were conducted in three laboratories, the process of transferring contextual functions appears robust.

All participants completed the sequence within 164 trials or less. The teaching procedure used in the present study reduced the number of trials per phase and the number of training phases, as compared to the procedure used by Pérez-González and Serna (2003). A hypothetical participant performing with no errors would need 368 trials to complete the sequence used by Pérez-González and Serna. In the present research, all participants completed the entire sequence in less than one third of those trials. The procedure was also very effective as compared to those that showed contextual control after hours of teaching and repeated testing (e.g., Meehan & Fields, 1995), but these authors showed transfer of control with one single contextual stimulus over new equivalence classes-not just a conditional discrimination.

Study 2

Study 2 examined how the procedure for teaching the functions of the contextual stimuli affects acquisition and transfer of contextual functions. Specifically, we analyzed the effect of introducing the two contextual stimuli interspersed in the first XAB teaching phase. Pilot studies in our lab showed that, when the two stimuli X1 and X2 were introduced intermixed, some participants did not respond correctly to the XEF test.

Method

Four participants received the same teaching procedure as in Study 1, except that Phases 4 (X1AB) and 5 (X2AB) were omitted from the teaching sequence (see Table 1). Thus, when finishing Phase 3, the participants passed directly to the XAB second-order conditional discrimination of Phase 6. In this phase, contextual stimuli X1 and X2 appeared interspersed across trials.

Results

Participants required from 201 to 279 trials to complete the entire sequence of phases. They made from 31 to 83 errors. The participants made a minimum of 15 errors in the first phase with XAB (Phase 6). Vera made 18 errors, Lora and Deba made 15 errors, and Vana made 65 errors. They learned the phases with XAB (Phases 6, 7, & 8) within 117 to 200 trials.

The participants' selections across the 24 XEF test trials are shown in Table 3. Vera responded correctly on each of the 24 trials of the XEF test. Lora responded correctly on 22 trials. Thus, these 2 participants acquired contextual control and transferred it to a new conditional discrimination. Deba showed variable responding in the session: (a) She chose among the F comparisons conditionally in the presence of both the contextual stimulus X and the sample E, but incorrectly (see Trials 1 and 3 in Table 3); (b) she responded conditionally on the X1 or X2 stimulus only. For example, she selected F1 given X1 and F2 given X2 (Trial 2 in Table 3); (c) she responded conditionally on the E stimuli only; for example, she selected F1 given E1 and F2 given E2, regardless of the X stimulus (Trial 4 in Table 3); and (d) she responded correctly on the last six trials (see columns for Trials 5 and 6 in Table 3). Vana generally selected F1 in the presence of E1 and F2 in the presence of E2, regardless of the X stimuli.

Discussion

Only 2 of 4 participants demonstrated acquisition and transfer of contextual functions to a novel conditional discrimination. Although the procedure of Study 2 did produce transfer of contextual functions (in 2 participants only), it appears to be somewhat less effective than the procedure used in Study 1 regarding transfer of contextual control. These results suggest that the introduction of X1 and X2 in two separate phases may facilitate acquisition and transfer of the contextual functions of X stimuli to novel conditional discriminations.

Two factors, at least, may be implicated in the transfer of contextual functions. One factor could be that the specific functions of X1 and X2 are better acquired when each stimulus is presented alone. Actually, behavior in the presence of stimulus X1 is the same as in its absence. Thus, participants can easily learn the function of stimulus X1 when this stimulus is presented alone in one phase, especially if the stimulus is introduced right after mastering the AB conditional discrimination. For similar reasons, the function of stimulus X2, which consists of selecting the comparison opposite to that chosen in its absence, could also be better learned if stimulus X2 is the only contextual stimulus in a phase. This hypothesis is consistent with the errors of participants of Study 1 in Phase 5, in which stimulus X2 was introduced. All participants made errors in the first trial.

The second possible factor is that the procedure used in Study 1 induced learning with very few errors. Participants of Study 1 made from 7 to 11 errors across all teaching phases. In contrast, the 2 participants of Study 2 who did not demonstrate transfer of contextual functions made 40 and 83 errors. It may be that learning with few errors facilitates the transfer of contextual functions.

The procedure of Study 1 had two more phases, with XAB, than the procedure of Study 2. It could be possible that the procedure of Study 1 was more effective in producing transfer of contextual functions than the procedure of Study 2 because the participants of Study 1 received more learning trials, or more reinforced trials, in the XAB phases than participants in Study 2. This did not happen. On the contrary, the 3 participants of Study 1 received only from 52 to 60 trials. The 4 participants of Study 2 received 102 trials or more, of which 84 trials or more were reinforced. Thus, number of trials per se, or number of reinforced trials per se, does not facilitate the transfer of contextual functions.

Deba responded correctly to the last six trials of the XEF test. Perhaps she would have kept responding correctly if the test had been extended beyond 24 trials; in this case, her results would be the same as those of the two participants who responded correctly. Thus, she would acquire and transfer contextual functions, but with some delay.

Participant Vana ignored the contextual stimuli in the XEF test. That responding can be explained by two possibilities: The first possibility is that stimuli X1 and X2 did not acquire contextual functions. In that case, the functions of the X stimuli would be similar to the functions of the A stimuli. The second possibility is that stimuli X1 and X2 had acquired contextual functions during the XAB teaching, but these functions did not transfer to the EF conditional discrimination. In the absence of further testing, it is impossible to know which of these two possibilities hold. A comparison between Studies 1 and 2, however, may give some hints to answer this question: Studies 1 and 2 differed only in the specific procedures to teach XAB. All participants of Study 1 responded correctly to the XEF test; in contrast, only 2 of 4 participants of Study 2 responded correctly. This suggests that the procedure of Study 1 was more effective than that of Study 2 to produce acquisition of contextual functions.

The results of Study 2 contrast with those obtained in other studies (Carpentier et al., 2002, 2003; Serna & Pérez-Gonzâlez, 2003). In those studies, all participants demonstrated transfer after the introduction of the contextual stimuli interspersed in a block of trials. The discrepancies between the results of the present research and those of previous studies may have been caused by other variations in procedures. For example, Carpentier et al. taught the contextual functions in a phase in which trials of two first-order conditional discriminations were intermixed with secondorder conditional-discrimination trials (analogous to intermixing AB, EF, and XAB). This manipulation could have facilitated the acquisition of functions of the contextual stimuli by facilitating discrimination between the contextual stimuli and between these stimuli and the absence of them. Another variation in the procedure that could influence the results concerns instructions. In the present studies, we avoided explaining the details of the task to the participants, with the purpose of minimizing participants' verbalizations.

Study 3

Study 2 showed that the way in which the X stimuli were presented affected acquisition or transfer of the contextual functions. Study 3 analyzed whether differential consequences are necessary during the teaching of the X stimuli. We examined in Study 3 whether presenting stimuli X1 and X2 without providing differential consequences would affect acquisition and transfer of contextual functions.

Method

Four participants received the same teaching procedure as participants of Study 1, except that the responses in phases with contextual stimuli did not have differential consequences. Thus, the participants received 12 trials of Phases 4 and 5 and 8 trials of Phases 6 to 8, regardless of their performance in these phases (see Table 1).

Results

Participants required from 142 to 295 trials to complete the entire sequence of phases. In this study, only trials of phases with the AB and EF conditional discriminations were administered with feedback (Phases 1 to 3 and 9 to 11). Sara made only five errors, all in Phase 1, with the AB conditional discrimination. Tera, Mona, and Joao made 20 or more errors; they made errors in the AB as well as in the EF conditional discriminations.

In the XAB phases, stimulus X1 or X2 were presented without feedback. Three participants (Sara, Tera, Mona) selected B1 conditionally on A1 and B2 conditionally on A2, independently of the X stimulus presented on each trial (with exceptions in less than five trials). The fourth participant's selections showed no apparent pattern.

In the XEF test, the 4 participants selected F1 in the presence of E1 and selected F2 in the presence of E2, with exceptions in only three trials, as shown in Table 4. Thus, the participants selected among comparisons according to the sample stimuli, independently of the presence of X1 and X2, just as they did when they learned the AB conditional discrimination.

Discussion

Responding in the XAB phases, with stimulus X1 or X2 presented without feedback, indicates whether the participants acquired contextual control. No participant selected conditionally on the X stimuli. Thus, none of the participants acquired contextual control. Their performance demonstrated that the mere exposure to stimuli X1 and X2, without differential consequences, was not sufficient to establish the contextual stimulus function. Thus, differential consequences during the presentation of the X stimuli appear to be necessary for the acquisition and transfer of contextual functions.

The participants ignored the X stimuli during the XEF test. This response pattern is consistent with the assumption that contextual functions were not acquired during the XAB phases. Thus, the results of Study 3 indicate that the presentation of the X stimuli in the XAB phase without feedback is not sufficient for the acquisition of contextual functions. Moreover, it is obvious that if contextual functions are not acquired, these functions cannot transfer.

：芭芭拉

Study 4

Studies by Carpentier et al. (2002) showed that when the X stimuli are introduced prior to the conditional discriminations, it is possible to get transfer of functions of contextual stimuli. Transfer, however, is less likely than when the conditional discrimination is taught first. In the study of Carpentier et al., it could have happened that the introduction of the X stimuli together with the conditional discrimination was sufficient for the X stimuli to acquire the functions of contextual stimuli in some participants. Study 4 explored whether the contextual functions of stimuli X can be acquired or may transfer without learning the initial conditional discrimination. Thus, we evaluated acquisition and transfer of contextual functions after teaching the second-order conditional discrimination directly.

Method

Three participants received the same teaching procedure as in Study 1, with the exception of Phases 1 to 3 (designed to teach the AB conditional discrimination) that were omitted. Therefore, the teaching procedure started when the X1AB phase was presented (Phase 4, see Table 1).

Results

Participant Joro required 180 trials to complete the sequence of phases of this study (he made 60 errors). Yola required 236 trials (she made 82 errors). Frao required 139 trials (he made 34 errors). They made most errors during the XAB phases.

The 3 participants failed to respond correctly during the XEF transfer test (see Table 5). Joro performed conditionally on the contextual stimuli, but not according to the definition of correct responding. Yola mostly selected F1 given E1 and F2 given E2, independently of the stimulus X that was present. Thus, Yola's selections demonstrated control by the sample stimuli (E1 and E2) and not the contextual stimuli (X1 and X2). Participant Frao did not show any evident pattern of selections among the comparisons.

Discussion

Performance of participants Yola and Frao showed that, even if participants learn the second-order conditional discriminations with X stimuli as contextual stimuli, contextual functions do not easily transfer to a new conditional discrimination when the AB conditional discrimination is omitted from the teaching procedure. Thus, these results suggest that the teaching of AB facilitates that the X stimuli acquire contextual functions.

Joro responded conditionally to stimuli X1 and X2 as well as stimuli E1 and E2. This performance indicates that stimuli X1 and X2 acquired functions separated from stimuli A1 and A2 during the XAB teaching. These functions transferred to the EF conditional discriminations. In the absence of the AB teaching, however, the functions of X1 and X2 could not be specific: In Studies 1 and 2, X1 indicated selections of the same comparison as in its absence; X2 indicated selections of the alternative comparison as in its absence. In the present study, the AB conditional discrimination was not taught without the X stimuli. Thus, it was impossible for stimuli X1 and X2 to acquire specific contextual functions. Instead, stimuli X1 and X2 could acquire a more generalized function: Indicating selections of one comparison given a sample in the presence of one stimulus X and the function of indicating selections of the alternative comparison in the presence of the other stimulus X. Thus, stimulus X1 could indicate selection of the same comparison as in its absence or selection of the other comparison as in its absence. Stimulus X2 should indicate the function alternative to that of stimulus X1. Indeed, in the XEF test, Joro selected in the presence of X2 the same comparison as in its absence, and in the presence of X1 he selected the alternative comparison; that was just one of the two possibilities. Serna and PérezGonzâlez (2003) observed this performance under other conditions and defined it as generalized contextual control.

According to this rationale, half of the participants who acquired generalized contextual control would respond as Joro did in the present study. The other half would respond as it was defined as correct (i.e., as participants of Study 1 did). One implication of this analysis is that responding "correctly" to the XEF test would not indicate transfer of specific functions. To determine whether transfer of specific functions or generalized contextual control occurred, it is necessary to conduct additional tests with novel conditional discriminations. The present study showed that, when the first-order conditional discrimination is not taught, stimuli X1 and X2 can acquire generalized contextual functions. They, however, cannot acquire specific contextual functions.

The results of the present study are consistent with those from Carpentier et al. (2002). They found less transfer when the contextual stimuli were introduced in the first phase than when the contextual stimuli were introduced after teaching the first-order conditional discriminations. The results of the present study are also consistent with those from Meehan and Fields (1995). In that study, only 3 out of 5 adults showed contextual control by a long bar (equivalent to X2) after between 10 and 54 hours of teaching and testing. They taught all the first order conditional discriminations with a short bar (equivalent to X1). It could be that the introduction of the stimulus equivalent to the X1 in a latter phase could have improved participant's performance when the stimulus equivalent to X2 was introduced for testing.

Study 5

Serna and Pérez-González (2003) showed that novel stimuli do not control conditional discriminations when they are presented prior to the establishment of the conditional discriminations. It is not known, however, whether transfer of contextual functions occurs when contextual stimuli are presented within novel conditional discriminations. Study 5 examined the effect of teaching the second conditional discrimination on the transfer of contextual stimulus functions to this conditional discrimination. Thus, the experimental question was to know whether, after learning a first conditional discrimination and bringing it under the contextual control of the X stimuli, the contextual functions of the X stimuli would transfer directly to a novel conditional discrimination.

Method

Four participants received the same teaching procedure as participants of Study 1, except that Phases 9 to 11 (designed to teach the EF conditional discrimination) were omitted. Thus, participants received the XEF test (Phase 12, see Table 1) immediately after learning the XAB conditional discriminations (Phase 8).

Results

In this study, only the phases with the AB and XAB conditional discriminations and the XEF test were administered. Elea required 111 trials to complete the entire sequence (she made 29 errors). Silva required 314 trials (she made 114 errors). Bela required 159 trials (she made 42 errors). Graa required 169 trials (she made 46 errors).

Performance in the XEF test is shown in Table 6. Participant Elea selected comparisons conditionally to the X and E stimuli. She consistently selected F2 given X1 and E1 or given X2 and E2, and selected F1 given X1 and E2 or given X2 and E1. In contrast to Study 1, X1 did not control E1-F1 and E2-F2 relations, and X2 did not control E1-F2 and E2-F1 relations.

Participants Sila and Bela always responded correctly (as defined above) on trials when stimuli X1-E2 were presented. Their responses in the presence of the other stimuli (e.g., X1-E1, X2-E1, and X2-E2), however, varied across trials. Sila's responses in the last 12 trials were controlled by contextual stimuli but not by the sample stimuli (she selected F2 given X1 and F1 given X2). Bera responded with a variable pattern in the presence of the other three stimulus combinations.

In contrast to Sila and Bela, Graa always responded incorrectly on trials when stimuli X1-E2 were presented. She responded with a variable pattern in the presence of the other three stimuli, just as Bela did.

Discussion

One participant (Elea) responded conditionally to the X stimuli and the sample. The other 3 participants showed variable responding. The participants received the same procedure for learning AB and XAB as participants in Study 1, a study in which the X stimuli did acquire contextual functions. Given this identity, it is reasonable to assume that the X stimuli also acquired contextual functions in Study 5. The contextual functions of the X stimuli, however, did not transfer in 3 participants. The 4 participants tended to behave according to the presence of the X and E stimuli. This response pattern was different from those exhibited by the participants in Studies 3 and 4. This fact may indicate that, after learning a contextually controlled conditional discrimination, there is a general disposition to respond according to the contextual stimulus and the sample in a new second-order conditional discrimination. This disposition, however, is not enough to guarantee transfer of functions to a novel conditional discrimination if that discrimination has not been learned previously.

Elea's performance in the XEF test was similar to the performance of Joro in Study 4. Stimulus X1 controlled E1-F2 and E2-F1 relations and stimulus X2 controlled E1-F1 and E2-F2 relations. Thus, as in Study 4, the contextual functions of stimuli X1 and X2 transferred. Contrary to Study 4, it is reasonable to assume that the specific functions of X1 and X2 were acquired (as explained above). In the present study, however, relations among the E and F stimuli were not taught. Therefore, the specific functions of stimuli X could not transfer. Thus, the present study indicates that, when the second first-order conditional discrimination is not taught, stimuli X1 and X2 may acquire specific contextual functions. Moreover, the contextual functions can transfer only in a generalized way when stimuli X1 and X2 are presented to novel stimuli that do not have preestablished relations among them.

The responding of Elea was also similar to that of the participants in Study 1 of Serna and Pérez-González (2003). In that study, 2 participants learned the AB and XAB conditional discriminations, as in the present study, and received trials of another, first-order conditional discrimination (CD) without feedback, until they behaved consistently. One participant selected D1 given C1 and D2 given C2; the other participant selected D2 given C1 and D1 given C2. In the transfer test for contextual functions, in the presence of X1 both participants chose the comparisons they had chosen in the CD task; these participants selected the other comparison in the presence of X2. Thus, these participants responded on the XCD test as Elea did in the XEF test of the present study. The forms of transfer in the study of Serna and Pérez-González and in the present study, however, differed. In Serna and Pérez-González's study, participants demonstrated transfer of specific functions. In the present study, Elea demonstrated generalized contextual transfer.

：芭芭拉

General Discussion

One goal of the present research was to replicate the findings reported by Pérez-González and Serna (2003). All participants from Study 1 learned the complete procedure with few errors and demonstrated transfer of contextual functions. These results suggest that all components of the learning sequence were sufficient for acquiring contextual functions and transfer the contextual functions to new conditional discriminations. The procedure used in the present research was very efficient, as the 3 participants completed the entire experimental sequence and showed transfer of specific functions within 164 trials or less (less than 13 minutes).

The main goal of the present research was to study the conditions under which the functions of contextual stimuli are acquired and transfer to novel conditional discriminations. For this purpose, we analyzed the role of the sequence training components in the acquisition and transfer of contextual control. In contrast with the performance of the participants of Study 1, none of the 11 participants who received an incomplete procedure (in Studies 3 to 5) showed transfer of specific contextual functions. Thus, the findings of these additional studies showed that the complete training sequence was necessary to establish contextual control and to transfer specific contextual control to novel conditional discriminations.

Furthermore, in Study 2 only 2 of the 4 participants who received a procedure with all conditional discriminations, but an XAB conditionaldiscrimination protocol less elaborate than in Study 1, demonstrated transfer. This suggests that teaching each contextual function separately plays a relevant role in the acquisition and transfer of contextual functions.

The present research shows several conditions that seem necessary for the acquisition and transfer of specific contextual functions. First, when we presented XAB without providing differential consequences, all participants failed to learn contextual control and, consequently, did not demonstrate transfer (Study 3). These results suggest that the presentation of two stimuli (X1 and X2) with differential feedback is necessary for these stimuli to acquire contextual functions over established relations and then to transfer these functions to new stimuli. The mere presence of two stimuli with the sample is not sufficient to exert control over the participants' selections.

Second, when the first first-order conditional discrimination (AB) was not taught, participants did not demonstrate the transfer of contextual functions (Study 4). These results suggest that the acquisition of a first-order conditional discrimination is necessary for the acquisition of contextual functions, as demonstrated in Study 4. This result is not surprising because when presented for the first time, stimuli X and A may have identical functions in controlling comparison selection (or they may function as a compound). For example, the simultaneous presentation of X1 and A1 control the selection of B1. Therefore, it is equally likely that X1 controls the A1-B1 relation, that A1 controls the X1-B1 relation, and that both stimuli control the selection of B1. In these conditions, the X stimuli might not acquire contextual functions over the relations among the remaining stimuli.

Third, when the second first-order conditional discrimination (EF) was not taught, participants did not demonstrate transfer of specific contextual functions (Study 5). These results suggest that the acquisition of this conditional discrimination is necessary for the transfer of the acquired functions, for reasons similar to those described above. In this case, however, the conditional discrimination is necessary to transfer, rather than to acquire, contextual functions.

In summary, the present results suggest that to obtain the transfer of specific functions of contextual stimuli, it is necessary (a) to teach a conditional discrimination, (b) to bring this conditional discrimination under contextual control, and (c) to teach a second conditional discrimination. In such conditions, the functions of the contextual stimuli in the first conditional discrimination transfer to the second conditional discrimination.

One participant of Study 4 and another participant of Study 5 selected comparisons during testing conditionally to the X stimuli and to the samples. They did not show, however, transfer of specific functions. Such performances indicate that when the first or the second first-order conditional discriminations (AB or EF) are not taught, it is not possible to transfer specific functions. Instead, transfer of generalized contextual functions may occur, as demonstrated also by Serna and Pérez-González (2003).

One might argue that the clear difference of performance between Study 1 and Studies 2 to 5 is explained by the fact that more trials were programmed in Study 1 than in the remaining studies. The number of trials would be the relevant variable, rather than teaching specific conditional discriminations or providing feedback in the XAB conditional discrimination. This possibility seems remote because of two facts: First, many studies in the area of stimulus equivalence have shown that teaching specific conditional discriminations, testing conditional discriminations, and manipulating order influence results greatly; meanwhile, the number of trials per se appears not to influence the results as much. Second, in the present research, most participants in Studies 2 to 5 actually received more learning trials than those of participants in Study 1. Specifically, all participants in Studies 2 to 5, except one, received more learning trials than Lara of Study 1-the participant who received fewer trials. Moreover, 11 of 15 participants of Studies 2 to 5 received more learning trials than the 3 participants of Study 1. Thus, these facts replicate previous results in that teaching specific conditional discriminations, rather than number of trials, determined performance. In the present research, teaching three conditional discriminations determined acquisition and transfer of contextual functions.

The present research analyzed factors involved in the establishment of a five-term contingency (Sidman, 1986) and the transfer of control exerted by the contextual stimuli (the fifth term). In a five-term contingency, reinforcement must be contingent on responding for selecting: (a) the appropriate comparison; (b) in the presence of a given sample; and (c), conditionally in the presence of a contextual stimulus. For appropriately establishing a five-term contingency, it is necessary that reinforcement be contingent upon comparison selection conditionally on the specific sample and contextual stimuli present in the trial. Across trials, the sample and the contextual stimuli must vary and the reinforced comparison must co-vary with them (Martínez & Moreno, 1995). In Study 3 of the present research, reinforcement was never made contingent upon the presence of the X stimuli. In conditional discriminations AB and EF, reinforcement for comparison selection depended upon the sample presented on each trial. Therefore, this study involved four-term contingencies, instead of five-term contingencies. The lack of transfer in the XEF test observed in Study 3 suggests that establishing five-term contingencies is necessary for learning and transferring contextual control. The results of Studies 4 and 5 suggest that teaching the two first-order conditional discriminations is necessary for the transfer of contextual control. Because contextual control in the present preparation refers to preestablished relations among samples and comparisons in conditional discriminations, it seems reasonable that teaching those two conditional discriminations be necessary for transfer. A key feature is that the relations established in the two conditional discriminations are of the same kind (e.g., matching to sample). Another scenario is possible, however, in which it is not necessary to teach the relevant relations. This is the case of having perceptual, rather than arbitrary, stimulus relations. In such circumstances, the contextual function may be taught with the perceptual relation established among the stimulus within a set. The contextual stimuli are then tested with another stimulus set involving the same perceptual relations. Obtaining transfer of contextual functions seems very likely. For example, the contextual functions may be taught with identity matching to sample on the A stimuli, and tested with identity matching to sample on the B stimuli. People would probably select among the B stimuli according to the sample and to the contextual stimuli. In a related study, for example, people learned the AA-X relation and showed transfer of function of comparison stimuli by appropriate choices on the BB-X relation (Pérez-González, 1991).

Two participants in the experiment by Serna and Pérez-González (2003) and 1 participant in our Study 5 behaved conditionally on the contextual stimuli and the sample even though the second conditional discrimination was not taught. In Serna and Pérez-González's experiment, this may have happened because arbitrary selection was very similar to standard matching to sample. In Study 5, conditional choices may occur because the participants find it easy to learn the relations of the second conditional discrimination (e.g., assigning F1 to E1) without experimental intervention.

Sidman's (1986, 1994, 2000) parsimonious analysis of contextual control could be useful in part to interpret the present data. The contextual stimuli did not control specific stimulus classes, because the contextual stimuli were presented with stimuli that did not appear before with the contextual stimuli. Still, we observed transfer. Rather than controlling specific stimulus classes, contextual stimuli controlled generalized responding. Contextual stimulus X1 might control the selection of stimuli of the same class, and contextual stimulus X2 might control the selection of stimuli of a different class (assuming that the relation directly taught is enough to consider these two stimuli as members of the same class). But an important feature of the present study is that contextual stimuli controlled selection in a conditional discrimination with novel stimulus classes that had not been directly or indirectly related to previous contextual stimuli. Sidman (1994, p. 548) suggested that a stimulus can belong to the same class as the relation established between two other stimuli. In the present case, however, the relation should be established between the contextual stimulus, on the one hand, and the sample and the correct comparison, on the other hand. Thus this analysis becomes complicated. Considering that the contextual stimuli do not belong to the same class as that of the remaining stimuli seems more parsimonious.

Sidman (1994) also affirmed that the term "transfer of function" is unnecessary because it merely denotes stimulus equivalence (for example, two stimuli are in the same class if they have the same functions). In the present case, however, it is appropriate to refer to transfer of functions because the functions of the X stimuli transferred even though the X stimuli did not belong to a stimulus class. Thus, the present results document the transfer of functions of stimuli X but not stimulus classes involving these stimuli.

The functions of contextual stimuli seem similar to the functions of instructions (for discussions of this analogy see Hayes, Barnes-Holmes, & Roche, 2001, p. 32, and Saunders, 1996). A key finding of studies about instructional control is that a correspondence between an instruction and the subsequent contingency must exist in order to get generalized instructional control (e.g., Baron & Galizio, 1983; Martínez-Sanchez & Ribes-lñesta, 1996). The functions of the contextual stimuli in the present research also consist of specifying subsequent relations. The present research shows that, for the contextual stimulus to acquire a particular function, a correspondence between the stimulus and the contingencies must exist, in the sense of determining the sample-comparison relations that will be reinforced (as demonstrated in Study 1). When there was no correspondence between the presence of the contextual stimuli and the first-order conditional discrimination, stimuli X1 and X2 appeared to be irrelevant and the participants ignored them (Study 3). For example, the functions of X1 may be similar to the instruction "Select the name of the person whose discipline is identical to that of the person named in the sample" in the example of Figure 1. In this case, to obtain the reinforcer one must select the name of the individual whose discipline is identical to that of the person named in the sample. The functions of X2 may be similar to the instruction "Select the name of the person whose discipline differs from that of the person named in the sample." In that case, to obtain the reinforcer one must select the name of the individual whose discipline differs from that of the person named in the sample.

The results of the present studies suggest that to follow these instructions, three things are necessary. First, the person must learn the relations between the names of the taught relations (i.e., between Cervantes and Tolstoi). Here the only relevant factor is that these figures share the same discipline-it is not necessary to know which discipline the two figures share. The necessity of teaching this relation was demonstrated in Study 4. Second, that person must learn the correspondence between the instructions ("Select the name of the person whose discipline is identical to that of the person named in the sample") and the following consequences. This necessity was demonstrated in Study 3. Third, that person must learn the relations between new names (e.g., that Bernard and Cajal shared the same discipline). The necessity of this last component was demonstrated in Study 5. Learning all these components is necessary for behaving according to the instructions "select the name with the same discipline" and "select the name with a different discipline" in the novel situation. The person who learned these components will follow the new rules: That person would select a name according to the instruction in the presence of the new names.

：芭芭拉

[Reference]
References
BARON, A., & GALIZIO, M. (1983). Instructional control of human O p e r ant behavior. The Psychological Record, 33, 495-520.
BUSH, K. M., SIDMAN, M., & DE ROSE, T. (1-9-8-9). Contextual control of emergent equivalence relations. Journal of the Experimental Analysis of Behavior, 51, 29-45.
CARPENTIER, F., SMEETS, P. M., & BARNES-HOLMES, D. (2002). Class formation of unrelated stimuli with same discriminative functions. European Journal of Behavior Analysis, 3, 7-19.
CARPENTIER, F., SMEETS, P. M., & BARNES-HOLMES, D. (2003). Matching unrelated stimuli with same discriminative functions: Training order effects. Behavioural Processes, 60, 215-226.
DYMOND, S., & BARNES, D. (1995). A transformation of self-discrimination response functions in accordance with the arbitrarily applicable relations of sameness, more than, and less than. Journal of the Experimental Analysis of Behavior, 64, 163-184.
GATCH, M. B., & OSBORNE, J. G. (1-9-8-9). Transfer of contextual stimulus function via equivalence class development. Journal of the Experimental Analysis of Behavior, 51, 369-378.
HAYES, S. C., BARNES-HOLMES, D., & ROCHE, B. (2001). Relational frame theory: A post-Skinnerian account of human language and cognition. New York: Plenum Publishers.
HAYES, S., KOHLENBERG, B. S., & HAYES, L. J. (1991). The transfer of specific and general consequential functions through sim-ple and conditional equivalence relations. Journal of the Experimental Analysis of Behavior, 56, 119-137.
KENNEDY, C. H., & LAITINEN, R. (1988). Second-order conditional control of symmetric and transitive stimulus relations: The influence of order effects. The Psychological Record, 38, 437-446.
LYNCH, D. C., & GREEN, G. (1991). Development and crossmodal transfer of contextual control of emergent stimulus relations. Journal of the Experimental Analysis of behavior, 56, 139-154.
MARKHAM, M. R., & DOUGHER, M. J. (1993). Compound stimuli in emergent stimulus relations: Extending the scope of stimulus equivalence. Journal of the Experimental Analysis of Behavior, 60, 529-542.
MARTÍNEZ, H., & MORENO, R. (1995). El papel de la contigüidad y la contingencia en el condicionamiento: Una revisión y una propuesta metodológica [The role of contiguity and contingency in conditioning: A review and a methodological proposal]. Mexican Journal of Behavior Analysis, 21 (Special issue), 75-106.
MARTÍNEZ-SÁNCHEZ, H., & RIBES-IÑESTA, E. (1996). Interactions of contingencies and instructional history on conditional discrimination. The Psychological Record, 46, 301-318.
MCILVANE, W. J., & DUBE, W. V. (1992). Stimulus control shaping and stimulus control topographies. The Behavior Analyst, 15, 89-94.
MEEHAN, E. F., & FIELDS, L. (1995). Contextual control of new equivalence classes. The Psychological Record, 45, 165-182.
PÉREZ-GONZÁLEZ, L. A. (1991). El análisis funcional de la conducta verbal a través del condicionamiento O p e r ante: La emergencia de nuevas conductas por medio de procedimientos de discriminaciones condicionales [The functional analysis of verbal behavior through O p e r ant conditioning: The emergence of novel behaviors by conditional discrimination procedures]. Oviedo, Spain: Servicio de Publicaciones de la Universidad de Oviedo.
PÉREZ-GONZÁLEZ, L. A., & SERNA, R. W. (1993a). Basic stimulus control functions in the five-term contingency. Experimental Analysis of Human Behavior Bulletin, 11, 52-54.
PÉREZ-GONZÁLEZ, L. A., & SERNA, R. W. (1993b, September). Comparación entre dos procedimientos de aprendizaje de estímulos contextuales en discriminaciones condicionales [Comparison between two procedures to learn contextual stimuli in conditional discriminations]. Communication presented to the V Meeting of the Spanish Society of Comparative Psychology. Barcelona, Spain.
PÉREZ-GONZÁLEZ, L. A., & SERNA, R. W. (2003). Transfer of specific contextual functions to novel conditional discriminations. Journal of the Experimental Analysis of Behavior, 79, 395-408.
PÉREZ-GONZÁLEZ, L A., SPRADLIN, J. E., & SAUNDERS, K. J. (2000). Learning-set outcome in second-order conditional discriminations. The Psychological Record, 50, 429-442.
ROCHE, B., & BARNES, D. (1996). Arbitrarily applicable relational responding and s-e-xual categorization: A critical test of the derived difference relation. The Psychological Record, 46, 451 -475.
ROCHE, B., & BARNES, D. (1997). A transformation of respondently conditioned stimulus function in accordance with arbitrarily applicable relations. Journal of the Experimental Analysis of Behavior, 67, 275-301.
SAUNDERS, R. R. (1996). From review to commentary on Roche and Barnes: Toward a better understanding of equivalence in the context of Relational Frame Theory. The Psychological Record, 46, 477-487.
SERNA, R. W. (1991). Interchangeably of stimulus terms in five-term contingencies. Experimental Analysis of Human Behavior Bulletin, 9, 2-3.
SERNA, R. W., & PÉREZ-GONZÁLEZ, L. A. (1994, October). Emergent five-term contingency performances. Communication presented to the Eleventh Annual Meeting of the Southeastern Association for Behavior Analysis, Savannah, Georgia, United States.
SERNA, R. W., & PÉREZ-GONZÁLEZ, L. A. (1997, July). Generalized contextual control of conditional discriminations and unreinforced conditional selection. Communication presented to the III European Meeting on Experimental Analysis of Behaviour. Dublin, Ireland.
SERNA, R. W., & PÉREZ-GONZÁLEZ, L. A. (2003). An analysis of generalized contextual control of conditional discriminations. Journal of the Experimental Analysis of Behavior, 79, 383-393.
SIDMAN, M. (1986). Functional analysis of emergent verbal classes. In T. Thompson & M. D. Zeiler (Eds.), Analysis and integration of behavioral units (pp. 213-245). Hillsdale, NJ: Lawrence Erlbaum.
SIDMAN, M. (1994). Equivalence relations and behavior: A research story. Boston: Authors CoO p e r ative.
SIDMAN, M. (2000). Equivalence relations and the reinforcement contingency. Journal of the Experimental Analysis of Behavior, 74, 127-146.
TOUCHETTE, P. E. (1971). Transfer of stimulus control: Measuring the moment of transfer. Journal of the Experimental Analysis of Behavior, 15, 347-354.
WULFERT, E., & HAYES, S. C. (1988). Transfer of a conditional ordering response through conditional equivalence classes. Journal of the Experimental Analysis of Behavior, 50, 125-144.

[Author Affiliation]
LUIS ANTONIO PÉREZ-GONZÁLEZ
University of Oviedo, Spain
HÉCTOR MARTÍNEZ
University of Guadalajara, México

[Author Affiliation]
This research was conducted under the Programa de CoO p e r atión Interuniversitaria America Latina-España 2001, from the Agenda Española de CoO p e r atión International, and by MCYT Grant BSO2002-00494 of the Spanish administration to Hector Martlnez, and under the Convention between the University of Guadalajara, Mexico, and the University of Oviedo, Spain, 2002. We thank Joseph Spradlin for comments and ideas, and François Tonneau and Gladys Williams for reviewing previous versions of the manuscript. Address correspondence to Luis Antonio Pérez-González, Despacho 219, Departamento de Psicología, Plaza Feijoo s/n, 33003 Oviedo, Spain (E-mail: laperez@uniovi.es) or to Héctor Martinez, lnstituto de Neurociencias, Francisco de Quevedo # 180 Col. Arcos Vallarta, C.P. 44100, Guadalajara, Jal., Mexico (E-mail: hectorm@cencar.udg.mx).