The role of discriminative stimuli on response patterns in resurgence.
Procedures that characteristically produce resurgence involve a minimum of three phases. (1) Beginning with the training phase, an initial response is acquired by way of reinforcement. In the following alternative-reinforcement phase, an alternative-response is reinforced and the initial response is placed on extinction resulting in its attenuation. Depending on the research desiderata, the said phase may be repeated to supply the number of alternative-responses required by the study. In the terminal resurgence phase, the prevailing operant contingency for the alternative-response(s) is modified such that the transient recovery of the initial response and its subsequent decrement are observed. Although extinction of the last-acquired alternative response has been widely employed in most studies, the operant parameters have expanded to include schedule-thinning (Lieving & Lattal, 2003), punishment (Wilson & Hayes, 1996), and progressive delay of alternative reinforcement (Jarmolowicz & Lattal, 2014). Taken together, the evidence suggests conventional extinction is not sine qua non of resurgence. Rather, resurgence is observed under a variety of disruptive factors to the prevailing reinforcement contingency with respect to the most recently acquired response.
Since the resurgence preparation permits experimental manipulations on selected aspects of behavioral history (i.e., configurations imposed on the training and alternativere-inforcement phases) and subsequently testing for response recovery in the resurgence phase, the procedure is well-suited for investigations pertaining to recall performance. One notable area of this application is in serial positioning, about which sequence responses are acquired along an ordinal series, and a recall test at a delayed time examines retention performance (Wright, 2007, 1998). The abovementioned list tasks have generally produced characteristic U-shaped patterns with superior recall performance for the first learned (primacy) or last learned items (recency) than for the middle learned items (intermediate effect). Studies have shown generality of the serial position effect across human and nonhuman species (Wright, 2007), stimulus dimensions (Knoedler, Hellwig, & Neath, 1999) and differential list lengths (Wright & Rivera, 1997). The resurgence procedure that involves training and one or more alternative-reinforcement phases are analogous to the list acquisition condition utilized in serial positioning studies. Likewise, the resurgence phase serves as the recall test condition wherein previous trained responses are observed.
Several studies have utilized the resurgence procedure to investigate serial positioning effects (Reid, 1994; Mechner & Jones, 2011). Notably, Reed and Morgan (2006) trained six rats to emit three different response sequences (e.g., left-left-right lever press) in succession using one training and two alternative-reinforcement phases. Next, the proportion of recovery among the three sequences was examined in two consecutive resurgence phases under conventional extinction. The results from the first resurgence phase showed recency effect for all rats, followed by the first response sequence as the next highest proportion for half of the rats, while two rats exhibited the greater proportion of the second response sequence and one rat showed equal proportion of the first and second response sequences. In the second resurgence test, all rats showed the greatest proportion of second response sequence, demonstrating an intermediate effect. The orderly recovery was such that upon the decrement of the most recently acquired response sequence, "the tendency is for the earlier of the other two trained response sequences to reemerge, followed by the other response sequences trained in the middle of the serial training program" (p. 313). The evidence supplied from previous studies suggests the training order of responses effect the number of responses recovered in extinction.
Although previous studies examining the ways with which response patterns are organized in resurgence have emphasized on the manipulation of operant parameters, the related research on operant renewal have directly examined response recovery upon changes to the extinction context (Bouton, Winterbauer, & Todd, 2012). In ABA and ABC renewal, the target response is reinforced in context (A), subsequently extinguished in novel context (B), and the response is recovered upon the reintroduction of the initial context (A) or novel context (C). AAB renewal involves reinforcement and subsequent extinction in the same context (A), and the test phase is conducted in a novel context (B). Experimentally, the contextual features manipulated in operant renewal have largely involved exteroceptive stimuli such as key lights (Podelsnik & Kelley, 2014), odor (Bouton, Todd, Vurbic, & Winterbauer, 2011), and stripes on chamber walls (Sweeney & Shahan, 2015).
In general, renewal studies involving three components have focused on altering the contextual variables while maintaining similar contingency arrangements across phases (i.e., reinforcement, extinction, and extinction). On the other hand, resurgence studies have largely examined the effects changing contingencies have on response recovery. Recent studies have involved conventional renewal in addition to resurgence preparation that consisted of systematic changes to the context in which responses were acquired and extinguished. In Podlesnik and Kelley's study (2014), pigeons were subjected to three groups: typical resurgence, modified resurgence, and renewal. Each group utilized a three-phase procedure in which the initial target response was acquired under one context using a specific key light configuration. In the second phase, a different key light configuration serving as contextual change was applied for all groups. In the typical and modified resurgence groups, the alternative response was reinforced and the target behavior underwent extinction, and no alternative reinforcement was scheduled in the renewal group. In the third phase, extinction was applied to all responses, and the context maintained for the typical resurgence group. However, the context returned to that of phase 1 for the modified resurgence and renewal group. The results showed greater resurgence in the modified resurgence group and more persistence of the alternative response in the typical resurgence phase. The enhanced resurgence obtained by the return to the initially trained context was also reported by Kincaid, Lattal, and Spence (2015). In their training phase, a concurrent resurgence procedure operated different schedules on two available keys. In the second phase, differential reinforcement-of-other behavior (DRO) was employed for both key pecks while one of the keys changed the light. In the resurgence phase, extinction was imposed on the responses and the renewal key returned to the same configuration found in the first phase.
Although these studies have examined the changes with context and operant schedules on two responses from which only the first acquired response served as the subject of resurgence, the present study aimed to examine the relative resurgence of two responses by utilizing a four-phase preparation that arranged for the stimulus conditions found in training and alternative-reinforcement phases and systematically presented them in the resurgence phase. From this approach, we attempted to extend upon prior studies on recall by showing the role of contextual control that alters the conventional serial position effect such that the middle acquired response results in the highest recovery in the resurgence test. Additionally, we examined the relative magnitude of resurgence with respect to context associated with reinforcement and/or extinction. While most resurgence studies have focused exclusively on the recovery of decremented responses, the present study reports on the emission patterns of untrained responses in the resurgence phase.
Approval from the University of Nevada, Reno institutional review board was obtained (Protocol Number: El 2-70), and 26 undergraduate students recruited from the psychology program completed this study. Eleven participants were male and 15 were female, and they were provided with extra credit for their psychology course as compensation. At any time of the experiment, the participants were permitted to withdraw without penalty. All participants were screened for color deficiency prior to the experiment. The maximum experimental session time was 50 minutes composed of informed consent, instruction and debriefing.
Experiments were held in a computer room that housed a 22" X 35" cubicle desk which served as the platform for a 17" monitor, keyboard, and mouse. A computer installed with the experimental program operated the events and data recording. A modified opaque, polyurethane typing mask overlaid the entire keyboard except for the following keys exposed on the numeric keypad: 1, 3, 5, 7, 9, "Delete" and "Enter." Additionally, the experimental program prohibited any entries other than the exposed keys. The mouse function was inoperative during the experimental events, and only applicable to initiate and terminate the program upon session completion.
Prior to entering the experimental chamber, the participants were provided with an inform consent form with the following description:
If you agree to participate in this study, you will be asked to complete a simple test to screen for color vision deficiency (color blindness). Upon completion you will be asked to participate in a computerized simulation for password entry. Instructions for the tasks will be presented via computer monitor. It is anticipated that this entire study will take between 10 and 60 minutes.
Subsequent to the completion of the consent form, the participants were led to the monitor station and screened for color deficiency. All participants included in this study passed the screening test.
The initial screen consisted of an instruction statement that described the response requirement, "Use the available keys to enter a password. A valid password must be 4 characters in length with no repeated characters." A "Proceed" icon located on the bottom right side of the screen initiated the experimental situation. Upon the termination of the experimental session, the monitor displayed a statement describing the session completion and to exit the computer room. For the purpose of examining extent of recurrence of three successively-trained response sequences, four experimental phases consisting of one training phase and two alternative-reinforcement phases were deployed before the presentation of the final resurgence phase.
Training Phase and Response Sequence Establishment
The session began with an instruction screen that displayed the statement, "Please enter your password," with a text input box located directly underneath. This display configuration was positioned approximately on the center of the screen. Since a single operandum (i.e., keyboard) was responsible for generating multiple response sequences, the parameters that established and demaracted each response sequence are as follows. In the reinforcement phase, the first entry that contained any four mutually exclusive characters (e.g., 5139), and submitted by pressing "Enter" effectively designated response-sequence 1. The establishment of such response sequence in this fashion fulfilled the unique configuration for response-sequence 1 from which other response sequences are differentiated. Response sequence in this phase that constituted any entries that deviated from response-sequence 1 (e.g., 4139) were considered as untrained response sequences and resulted in no program consequences. Deletion to entries prior to submitting with "Enter" was permitted, and such attempts and partial submissions (e.g., less than 4-keystrokes with "Enter") were not counted as proper response sequences.
The programmed consequences for response-sequence 1 consisted of a 2-s feedback screen composed of an oscillating expanding and shrinking star embedded with the text, "Congrats, the security system works!" on a fixed-ratio 1 (FR1) schedule. This was followed by a 2-s black out screen upon which the instruction screen was presented for the next trial. No programmed consequence followed untrained response sequences such that the inputted sequence was deleted upon entry and the instruction screen remained indefinitely until a valid entry had been made. The response rates were obtained from consecutive trial blocks composed of duration in seconds for three valid response sequence emissions (response-sequence 1 in this phase). In each trial block, all other response sequences (untrained response sequences in this phase) were recorded within the trial block until the completion of the third response-sequence 1 that initiated the next trial block. A stability criterion similar to the one described by Lieving and Lattal (2003) was used to determine phase transitions. The criterion specifies that in the final six trial blocks of a phase, the mean rate of the first three trial blocks and the mean rate of the last three trial blocks must be equal or less than 10 % of the grand mean rate of the six trial blocks. Continuous calculations of the response rates were performed in the background, and the performance fulfillment of the criterion resulted in transition to the next phase.
Alternative Reinforcement 1 Phase
In the second phase, the establishment of the response-sequence 2 was determined by the first entry in this phase that contained four mutually exclusive characters that differed from response-sequence 1 ascertained in the previous training phase. Response-sequence 2 was reinforced on a FR1 schedule, and response-sequence 1 emissions produced no feedback screen. All entries that differed from the character compositions of response-sequences 1 and 2 were considered as untrained response sequences such that they produced no programmed consequences. Phase change occurred when the rates of response-sequence 2 had achieved stability according to the abovementioned criterion.
Alternative Reinforcement 2 Phase
This aim of this phase was to generate a third unique response sequence that serves as the response-sequence 3 by way of the same arrangement employed in the previous phases. Response-sequence 3 was placed on a FR1 schedule for feedback screen presentation and no programmed consequences followed response-sequences 1 and 2 emissions. Response sequences that differed from 1, 2, and 3 were considered as untrained response sequences that resulted in no feedback screen. The phase transitioned once the rates of response-sequence 3 met the stability criteria.
In the final phase, conventional extinction for all three established response sequences was imposed by which the feedback screen was withheld. Additionally, untrained response sequences differed when the three established sequences produced no programmed consequences. Consecutive 15-s bins in lieu of trial blocks were used to observe the rate of occurrences with respect to the three established and untrained response sequences. The phase lasted for five minutes, and the entire experiment concluded at the end of this phase.
Background Color Grouping
The participants were randomly distributed into four groups to compare the performance with respect to stimuli correlated with reinforcement in the resurgence phase. Four colors--white, red, green and blue--were coordinated using the hue, saturation, and lightness (HSL) parameters to specify the color compositions. The colors occupied the entire screen as the background stimulus behind the instruction text and the input box, and they remain in effect until phase transition. For group 1, white (H: 360, S: 100 %, L: 100 %) was assigned as the background color in all four phases. The color assignments for groups 2, 3 and 4 utilized the same color assignments for the first three phases: red (H: 0, S: 100 %, L: 50 %) in training phase, blue (H: 240, S: 100 %, L: 50 %) in alternative reinforcement phase 1, and green (H: 120, S: 100 %, L: 50 %) in alternative reinforcement phase 2. In the resurgence phase, the background colors red, blue, and green were assigned to groups 2, 3, and 4, respectively. With these arrangements, the stimulus conditions were constant between third phase (alternative reinforcement phase 2) and resurgence phase for groups 1 and 4, such that the only change was the operant contingency.
Table 1 displays the number of participants, the mean rates and proportions for the for three trained and untrained response sequences found in all of the groups across the four experimental phases. Specific to the acquisition performance found in the training, alternative reinforcement phases 1 and 2, the corresponding reinforced response sequence held the highest mean rates across each group. The alternative reinforcement 1 phase observed the reduction of response-sequence 1 rates while the reinforced response-sequence 2 saw reliable rate increases across all groups. Alternative reinforcement 2 phase resulted in the rate reductions for response-sequences 1 and 2 while acquisition of response-sequence 3 was observed. Notably in this phase, all except for group 3 showed recovery of response-sequence 1 although these rates were lower compared to response-sequence 2 that underwent extinction. The mean proportions displayed in Table 1 included the untrained response sequence emissions such that the relative distributions of all response types were accounted for in each of the phases. The mean proportions of the three trained response sequences correspond to the differential reinforcement schedules found in the first three phases and the greatest distribution was observed for the respective reinforced response sequence.
All individual proportions of the three trained response sequences obtained in the resurgence phase are shown on Fig. 1. Unlike the aggregate proportions that included untrained response sequence emissions reported in Table 1, these measures were drawn exclusively from the trained response sequences to examine the relative level of the reinforced responses that underwent extinction. Across the groups, all except for one participant (96.2 %), demonstrated response-sequence 1, and response-sequence 2 was obtained for 21 participants (80.8 %). Notably, two participants from group 2 showed exclusive response-sequence 1 emissions. Taken together, resurgence as interpreted by the recurrence of previously acquired subsequently decremented responses (i.e., response-sequences 1 and 2) was observed with one or with both response sequences across all participants.
These data were subjected to a one-way ANOVA with 0.05 level of significance for each of the trained response sequence proportions obtained in the resurgence phase between the four groups. Response-sequence 1 analysis revealed statistically significant differences between groups, F(3, 22) = 3.88, p = .023. Protected t-tests performed on each of the pair-wise comparisons found statistically significant differences between group 2 and all other groups. Group 2 proportions were higher than group 1, t(12) = 2.89, p = .008, group 3, 1(10) = 2.61, p = .016, and group 4, t(12) = 21.71, p = .013. Comparisons were the remaining pairs, groups 1 and 3, t(10) = .03, p = .979, groups 1 and 4, t(12) = .18, p = .862, and groups 3 and 4, t(10) = .13, p = .895 revealed no statistically significant differences in response-sequence 1 proportions. Response-sequence 2 analysis confirmed statistically significant differences between groups, F(3, 22)= 12.27, p < .001. Protected t-tests on each pair-wise comparison showed group 3 proportions were higher than group 1, t(10) = 4.15, p < .001, group 2, t(10) = 5.88, p < .001, and group 4, t(10) = 4.62, p < .001. With respect to the remaining pairs, there was no statistically significant difference observed between groups 1 and 2, t(12) = 1.89, p = .071, groups 1 and 4, t(12) = .52, p = .610, and groups 2 and 4, t(12) = 1.38, p = . 182. Response-sequence 3 analysis indicated a statistically significant difference in the proportion between the groups, F(3, 22) = 5.40, p = .006. Protected t-tests on each pair-wise comparisons indicated group 4 proportions were greater than group 2, t(12) = 2.25, p = .035, and group 3, t(10) = 3.34, p = .002, but not with group 1, t(12) = .18, p = .856. Results of the remaining pairs revealed statistically significant difference between groups 1 and 3, t(10) = 3.26, p = .004, but no statistically significant differences were found between groups 2 and 3, t(10) = 1.37, p = .183. Differences between groups 1 and 2 fell just short of our employed level of significance, t(12) = 2.06, p = .051.
[FIGURE 1 OMITTED]
Group summaries displaying the medians and interquartile range for each of the response sequence emissions are depicted in Fig. 2. For groups 1 and 4, overall higher occurrences of response-sequence 1 and 3, relative to the emission of response-sequence 2 are observed. The u-shaped curve with prominent primacy and recency effects is consistent with the general patterns reported in serial positioning studies (Wright, 2007; Castro & Larsen, 1992). Group 2 median emissions revealed a similar serial positioning curve with the highest median obtained for the first acquired response-sequence 1 followed by the last acquired response-sequence 3, with the middle trained response-sequence 2 as the lowest median. Group 3 summary shows the highest median value obtained for the response-sequence 2, followed by response-sequence 1 and 3. Drawn together, they form an inverted U-shape with a pronounced intermediate effect that is in contrast with the characteristic serial position profile. From the patterns obtained from groups 2 and 3, the highest level of recovery involved the response sequence that corresponds to the presence of background color correlated with reinforcement. The highest emissions of response-sequence 3 found in group 4 suggests greater resistance to extinction when it was exposed for the first time in the resurgence phase.
Figure 3 shows the group cumulative records drawn from the resurgence phase and they were generated by summing the number of response sequence emissions for the current and the preceding intervals across all participants. Group 1 results show relatively consistent acceleration for response-sequence 3 followed by response-sequence 1, with the lowest acceleration for response-sequence 2. Group 2 performances show prominent runs for response-sequence 1 and the data path involves an accelerated burst from bins 11 to 20. In contrast, response-sequence 2 began with the second highest acceleration until bin 9 at which it was surpassed by response-sequence 3. In other words, response-sequence 3 prominence did not appear until response-sequence 2 ceased emissions between bins 10 and 13, while the emission of the last trained response increased from that point and persisted until the end of the session. This within-session inspection of group 2 revealed the eventual higher proportion of response-sequence 3 relative to 2 was not obtained by way of even distribution of emissions throughout the resurgence phase. Rather, the high emission for the last trained response did not appear until the middle part of the phase indicating a switch point of emissions at which the data paths were crossed. For group 3, the highest acceleration and sustainment were observed for the middle trained response-sequence 2. Out of the four groups, these participants emitted the lowest aggregated acceleration for the last trained response. The second highest path was response-sequence 1, with some brief pauses toward the latter half of the resurgence phase. Response-sequence 3, however, showed limited runs and an extended period of pauses (i.e., bins 9 to 18). Group 4 results show high acceleration for response-sequence 3, of which runs were sustained for the initial 14 bins suggesting the longest sustained resistance to extinction relative to other groups. Response-sequence 1 was steady but of lower acceleration relative to response-sequence 3, and this was followed by mostly abated runs for response-sequence 2.
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
As previously described, untrained response sequences accounts for non-reinforced keystroke sequences obtained in the four phases. The untrained response sequences in this respect are comparable to variable response sequences since their keystroke compositions were distinctive compared to those found in the three established response sequences. Extinction-induced variability is well documented in literature and found in nonhuman (Neuringer, Kornell, & Olufs, 2001) and human studies (Morgan & Lee, 1996), and they appeared as a notable feature with respect to resurgence. Although the mean rates and proportions of untrained response sequences were higher in the training phase, subsequent phases saw decreases before increasing as the most prominent response sequence in the resurgence phase for all four groups (Table 1). The aggregate cumulative records for the untrained responding obtained for participants in each group are displayed in Fig. 4. A least-squares linear trend line was fitted to each data path with high percentages of variance accounted for (VAF; range 99.2-99.9 %) such that the slopes were obtained for comparison. The order of the acceleration according to the slope value is as follows: group 1 (m = 29.6), group 4 (m = 21.8), group 2 (m = 19.4), and group 3 (m = 16.7). The absence of discriminative stimuli correlated with a specific response sequence and reinforcement found in group 1 yielded the highest emission of untrained response sequences. Whereas the presence of the discriminative stimulus in groups 2, 3, and 4 resulted in less emissions of untrained response sequences with greater allocation to the three trained response sequences.
[FIGURE 4 OMITTED]
While previous examinations in resurgence have maintained similar stimulus conditions in the training, alternative reinforcement, and resurgence phases, the present investigation revealed consistent patterns in which particular operants occur reliably in the presence of the discriminative stimulus relative to its absence in resurgence. In each of the colored background groups (2, 3, and 4), the highest proportion of response sequence observed corresponds to the respective discriminative stimulus while the no-color group 1 profile resemble ones found in conventional serial positioning effects (Wright, 2007). Notably, the highest proportion of the middle-trained response sequence recovered in group 3 yielded an intermediate effect of which is in contrast to the conventional serial positioning effects. While Reed and Morgan (2006) also obtained intermediate effects when the resurgence test under equal stimulus conditions was repeated, the present study shows the intermediate effect was achieved in the initial resurgence condition in the presence of the middle-trained discriminative stimulus.
The greater proportions of response-sequence 1 in group 2, and responses-sequence 2 in group 3 relative to those found in group 1 suggest renewal effects with which the return to the context associated with the respective response sequence acquisition enhanced the magnitude of resurgence. These results support previous findings by Kincaid, Lattal, & Spence (2015) in which ABA renewal produced higher recovery when one of the concurrently available keys returned to the color configuration used during the training phase of the response. However, their study incorporated a differential reinforcement-of-other behavior (DRO) schedule in the alternative reinforcement phase in contrast to our preparation that involved differential reinforcement of alternative behavior (DRA). Although DRO results in response decrement similar to the DRA effect on the target response, the absence of reinforcement of the alternative response may result in different resurgence patterns that makes direct comparisons difficult. Moreover, it is unclear whether the greater resurgence obtained in their study was due to absence of conventional extinction histories in the alternative reinforcement phase or the contextual change found in the test phase. The present study addressed those areas and extends upon the finding of increased resurgence by way of conventional extinction and returning to the context associated with reinforcement in the resurgence test.
Higher magnitude of resurgence when returning to reinforcement context was also found in Podlesnik and Kelley's (2014) modified resurgence procedure and the configurations used in groups 2 and 3 of the present study seemingly align with those employed their study. That is, conditions in resurgence return to the exact stimulus parameter found in one of the earlier phases. For Podlesnik and Kelly, the stimulus condition in the initial training phase consisted of an illuminated key for the target response with alternative response key darkened and inoperative. This arrangement was subsequently reapplied in the resurgence phase which amounts to the re-presentation of the discriminative stimulus associated with the initial target response acquisition. However, such configuration prohibited the alternative responding insofar that the results were interpreted with respect to stimulus control by way of the discriminative stimulus presentation. In contrast, the present study maintained the operandum without constraints to the type of responses that may be emitted. In light of the procedural differences, the present findings along with a growing number of studies suggest the reinforcement context increases the magnitude of response recovery.
In Sweeney and Shahan's (2015) study, they found that the magnitude of resurgence did not depend on whether alternative reinforcement was presented in context different than one found in the initial training phase. Our response-sequence 1 and 2 proportions did not yield significant differences between group 1 (alternative reinforcement in the same context) and group 4 (alternative reinforcement in different context with the same context across alternative reinforcement 2 and resurgence phase). However, alternative reinforcement was also presented in different contexts in groups 2 and 3, and significant differences of response-sequence 1 and 2 proportions were found relative to group 1. Taken together, our results support Sweeney and Shahan's observation that magnitude of resurgence does not depend on the same or different context in which alternative reinforcement was delivered. Rather, greater magnitude of resurgence depends on context associated reinforcement in the resurgence phase during which alternative reinforcement is removed. Future work may incorporate a novel context (no associations with reinforcement, alternative reinforcement, and explicit extinction) in the resurgence phase to examine the relative magnitude of recovery with respect to context associated with reinforcement.
The results of the present study suggest the removal of alternative reinforcement was sufficient to engender resurgence of response-sequence 1 and 2. However, the different recovery patterns with respect to contextual variables draw considerations from recent findings in operant renewal studies. According to the inhibitory context-response hypothesis (Todd, Vurbic, & Bouton, 2014), the extinction context (e.g., second of the three-phase renewal preparation) acquires discriminative function such that the response is not emitted when the context is present. Renewal effect is due in part when the extinction context is removed in the test phase (Bouton, Winterbauer, & Todd, 2012). Further support is found in the reduction of relative renewal when the response was tested in context associated with extinction (Todd, Vurbic, & Bouton, 2014, Experiment 3). Like some of the resurgence preparations previously described, our alternative reinforcement phases involved DRA such that the same context was associated with extinction for the preceding response sequence(s) and reinforcement of the current response sequence. If the response sequences are treated separately according to their context associations with the contingencies (reinforcement and alternative reinforcement with extinction) then the renewal account may help explain the relative magnitude of resurgence. With respect to group 1, response-sequence 1 and 2 received reinforcement and extinction associated with the same context. The context remained unchanged in the resurgence phase, and the relatively weak recovery may be attributed presence of the inhibitory extinction context. In contrast, group 2 that involved S1 (red) in the resurgence phase is analogous to removal of the extinction context since S1 was never associated with extinction with respect to response-sequence 1 and higher magnitude of recovery was observed. Similarly, response-sequence 2 also showed greatest recovery in group 3 in which S2 was never correlated with extinction with respect to the said response sequence. However, the removal of extinction context alone does not appear to engender greater magnitude without the presence of the context associated with reinforcement as observed in the weak recovery of response-sequence 2 in group 2. Taken together, the present results suggest the elimination of inhibitory extinction context promotes relative response recovery, but the increase of magnitude of recovery is largely due to the context associated with reinforcement. This is consistent with relative renewal strength with which return to the reinforcement context in ABA renewal produces greater recovery relative to extinction and novel contexts (Bouton, Todd, Vurbic, & Winterbauer, 2011; Todd, Vurbic, & Bouton, 2014).
The pattern of untrained responses that varied from the three acquired response sequences obtained in the present study is in accord with those found in the extinction and related resurgence literature (Morgan & Lee, 1996; Neuringer, Kornell, & Olufs, 2001; Mechner & Jones, 2011). Operant extinction not only increases the types of variability (e.g., topographical dimensions), but also the number of such responses in extinction (Neruinger, Kornell, & Olufs, 2001). Furthermore, various studies have shown variable responding can be brought under discriminative control (Denney & Neuringer, 1998; Ward, Kynaston, Bailey, & Odum, 2008). The overall proportion data (Table 1) found that untrained response sequences constituted the majority of responding that exceeded the three reinforced response sequences in all of the groups, with the highest mean proportion of untrained response sequence emission obtained in group 1. From the standpoint of stimulus control, it appears the presence of discriminative stimulus (groups 2,3,4) resulted in greater allocation of corresponding trained response sequence that which competed against the allocation of untrained response. Conversely for group 1, the absence of stimulus condition that correlates with the acquisition of a particular trained response sequence resulted in greater emissions of untrained response sequences.
One possible account lies in the procedure by which the second and third response sequences were established in the two alternative-reinforcement phases. Upon entering the alternative reinforcement phase, the reinforcement requirement was arranged for the first response sequence that was different than ones established in the previous phase (in the case of response-sequence 3, the requirement was the difference from the previous two response sequences). This is tantamount to a local lag differential reinforcement schedule (Page & Neuringer, 1985), as extinction is made contingent on the emission of the previous response sequence and remains in effect until a different response sequence that meets the schedule parameter in produced. Although this schedule was employed only initially in both alternative reinforcement phases until a proper alternative response has been established, nevertheless, a history of emitting variable response sequences was acquired before encountering the resurgence phase. The role of minimal exposure to a lag schedule on subsequent variable responding remains unclear without direct comparison. However, it was observed that group 1 had received the most exposures to this arrangement relative to other groups. This is by virtue that the stimulus condition was constant across all phases, and the global extinction in the resurgence phase resembled more closely to the local lag schedules in the preceding alternative reinforcement phases. As such, it is likely that the white background acquired the discriminative function associated with a larger degree variable responding history as opposed to the colored backgrounds. Future research might consider arranging for an explicit initial reinforcement phase in which a variable responding under a specific context is reinforced and subsequently tested for resurgence when a more recently acquired alternative response is placed on extinction.
In terms of relapse in treatments that utilize differential reinforcement contingencies (Higgins, Heil, & Lussier, 2004; Wacker et al., 2011; Wacker, Harding, Morgan, Berg, Schieltz, Lee, & Padilla, 2013), the resurgence models have largely conceptualized the problem behavior as the initial response that reemerges when the more recently acquired alternative response undergoes changes in the operant contingency. The present results revealed the challenges to relapse prevention by highlighting the generality of the primacy effect. By excluding the performance found in group 1 and 2, response-sequence 1 held the second highest proportion recovered in all of the groups. However, response competition by way of greater allocation to variable responding observed in group 1 may provide potential strategies to guide future treatment designs. Conceptually, if greater distribution is made to variables responding given stimulus conditions that have been associated with differential reinforcement of multiple response class members, then such potentiality of competition may attenuate the primacy effects when disrupters are made to the alternative response. Future research that translates the basic findings to clinical application will be fruitful to the growing literature on treatment relapse.
There are several limitations that warrant consideration. First, the participant size for group 3 (n = 5) was less than the other three groups (n = 7). One situation was due to subject attrition, in which the participant terminated the experiment prior to completion. For another participant, a technical malfunction prevented the session completion. In both cases, the preliminary data were discarded from the final analyses. Nevertheless, a general intermediate effect was observed and future studies should seek to expand and equate the group member size. Second, the amount of resurgence obtained was relatively low when compared to studies with nonhuman subjects. Potential procedural designs might mitigate this aspect by extending the training and alternative-reinforcement phases and increasing the number of responses emitted in the phase change criterion. Third, the lack of a dedicated renewal group in the present study prevents proper comparisons such that robust increase of response-sequence 2 in group 3 due to contextual influence is not fully known. The inclusion of a conventional renewal group in future studies utilizing similar four-phase preparations will provide more adequate evaluation of these differences. Finally, due to the exposure to extinction as part of the response sequence establishment (i.e., lag differential reinforcement), future research may seek to impose predetermined response dimensions (e.g., response sequences) such that unique response sequence generation will not occur on the part of the participant.
The present investigation supports the role of environmental determinants on recurrence of previously trained and attenuated behavior such that the presence of the discriminative stimulus in resurgence is positively correlated with the greater emission of the response that corresponds to the reinforcement history under the context. Furthermore, the four-phase preparation may contribute to future examination of more than one available response recovered in resurgence and permit additional contextual arrangements to address renewal effects. The present findings accentuate the importance of contextual variables with respect to the process of resurgence as well as the general pattern of which members of a response class recovers.
Author Notes We thank Richard W. Breuner for conducting the study, Pat Ghezzi and Neil Martin for their comments, and the anonymous reviewers who helped improve the quality of this paper. This study was supported in part by SEEK Education, Inc. and the contents are solely the responsibility of the authors.
Compliance with Ethical Standards
Conflict of Interest The authors declare that they have no conflict of interest.
Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.
Informed Consent Informed consent was obtained from all individual participants included in the study.
Bouton, M. E., Todd, T. P., Vurbic, D., & Winterbauer, N. E. (2011). Renewal after the extinction of free operant behavior. Learning & Behavior, 39, 57-67.
Bouton, M. E., Winterbauer, N. E., & Todd, T. P. (2012). Relapse processes after the extinction of instrumental learning: Renewal, resurgence, and reacquisition. Behavioural Processes, 90, 130-141.
Castro, C. A., & Larsen, T. (1992). Primacy and recency effects in nonhuman primates. Journal of Experimental Psychology, 18, 33-340.
Denney, J., & Neuringer, A. (1998). Behavioral variability is controlled by discriminative stimuli. Animal Learning and Behavior, 26, 154-162.
Epstein, R. (1983). Resurgence of previously reinforced behavior during extinction. Behaviour Analysis Letters, 3, 391-397.
Epstein, R. (1985). Extinction induced resurgence: Preliminary investigations and possible implications. Psychological Record, 35, 143-153.
Higgins, S. T., Heil, S. H., & Lussier, J. P. (2004). Clinical implications of reinforcement as determinant of substance use disorders. Annual Review of Psychology, 55, 431-461.
Jarmolowicz, D. P., & Lattal, K. A. (2014). Resurgence under delayed reinforcement. The Psychological Record, 64, 189-193.
Kincaid, S. L., Lattal, K. A., & Spence, J. (2015). Super-resurgence: ABA renewal increases resurgence. Behavioural Processes, 115, 70-73.
Knoedler, A. J., Hellwig, K. A., & Neath, I. (1999). The shift from recency to primacy with increasing delay. Journal of Experimental Psychology, 25, 474-487.
Lattal, K. A., & St. Peter Pipkin, C. (2009). Resurgence of previously reinforced responding: Research and application. The Behavior Analyst Today, 10, 254-265.
Lieving, G. A., & Lattal, K. A. (2003). Recency, repeatability, and reinforcer retrenchment: An experimental analysis of resurgence. Journal of the Experimental Analysis of Behavior, 80, 217-233.
Mechner, F., & Jones, L. (2001). Number of prior repetitions of operants, and resurgence. Retrieved from http://mechnerfoundation.org/pdf_ downloads/number_of_repetitions.pdf
Mechner, F., & Jones, L. (2011). Effects of sequential aspects of learning. Mexican Journal of Behavior Analysis, 37, 109-138.
Morgan, D. L., & Lee, K. (1996). Extinction-induced response variability in humans. The Psychological Record, 46, 145-159.
Neuringer, A., Kornell, N., & Olufs, M. (2001). Stability and variability in extinction. Journal of Experimental Psychology. Animal Behavior Processes, 27, 79-94.
Page, S., & Neuringer, A. (1985). Variability is an operant. Journal of Experimental Psychology. Animal Behavior Processes, 11, 429-452.
Podlesnik, C. A., & Kelley, M. E. (2014). Resurgence: Response competition, stimulus control, and reinforcer control. Journal of the Experimental Analysis of Behavior, 102, 231-240.
Reed, P., & Morgan, T. A. (2006). Resurgence of response sequences during extinction in rats shows a primacy effect. Journal of the Experimental Analysis of Behavior, 86, 307-315.
Reid, A. K. (1994). Learning new response sequences. Behavioural Processes, 32, 147-162.
Sweeney, M. M., & Shahan, T. A. (2015). Renewal, resurgence, and alternative reinforcement context. Behavioural Processes, 116, 43-49.
Todd, T. P., Vurbic, D., & Bouton, M. E. (2014a). Behavioral and neurobiological mechanisms of extinction in Pavlovian and instrumental learning. Neurobiology of Learning and Memory, 108, 52-64.
Todd, T. P., Vurbic, D., & Bouton, M. E. (2014b). Mechanisms of renewal after the extinction of discriminated operant behavior. Experimental Psychology: Animal Learning and Cognition., 40, 355-368.
Volkert, V. M., Lerman, D. C., Call, N. A., & Trosclair-Lasserre, N. (2009). An evaluation of resurgence during treatment with functional communication training. Journal of Applied Behavior Analysis, 42, 145-160.
Wacker, D. P., Harding, J. W., Berg, W. K., Lee, J. F., Schieltz, K. M., Padilla, Y. C., & Shahan, T. A. (2011). An evaluation of persistence of treatment effects during long-term treatment of destructive behavior. Journal of the Experimental Analysis of Behavior, 96, 261-282.
Wacker, D. P., Harding, J. W., Morgan, T. A., Berg, W. K., Schieltz, K. M., Lee, J. P., & Padilla. (2013). An evaluation of resurgence during functional communication training. The Psychological Record, 63, 3-20.
Ward, R. D., Kynaston, A. D., Bailey, E. M., & Odum, A. L. (2008). Discriminative control of variability: Effects of successive stimulus reversals. Behavioural Processes, 78, 17-24.
Wilson, K. G., & Hayes, S. C. (1996). Resurgence of derived stimulus relations. Journal of the Experimental Analysis of Behavior, 66, 267-281.
Wright, A. A. (1998). Auditory and visual serial positioning functions obey different laws. Psychonomic Bulletin & Review, 5, 564-584.
Wright, A. A. (2007). An experimental analysis of memory processing. Journal of the Experimental Analysis of Behavior, 88, 405-433.
Wright, A. A., & Rivera, J. J. (1997). Memory of auditory lists by rhesus monkeys (Macaca mulatto). Journal of Experimental Psychology. Animal Behavior Processes, 23, 1-9.
(1) Procedural variants include an explicit extinction phase (e.g., Epstein, 1983; Lieving & Lattal, 2003) in which the initial response is placed on extinction and response attenuation is obtained prior to the introduction of the alternative reinforcement phase. See Lattal & St. Peter Pipkin (2009) for further discussion.
James E. King  [ID] * Linda J. Hayes 
Published online: 7 March 2016
This paper has not been previously published, is not under review elsewhere, and will not be submitted elsewhere while it is under review by The Psychological Record.
[mail] James E. King
 Department of Psychology/296, University of Nevada, Reno, Reno, NY 89557, USA
Table 1 Mean response rates (emissions/minute) and overall proportions of emssions obtained for each response sequence types during the 4 phases across groups Training Phase RS1 RS2 RS3 Untrained Group 1 (n = 7) Mean Rate 11.48 -- -- 0.46 (SD) (5.83) -- -- (0.51) Mean Proportion 0.96 -- -- 0.04 (SD) (0.04) -- -- (0.04) Group 2 (n = 7) Mean Rate 7.97 -- -- 2.15 (SD) (2.82) -- -- (2.02) Mean Proportion 0.79 -- -- 0.21 (SD) (0.19) -- -- (0.19) Group 3 (n = 5) Mean Rate 9.12 -- -- 1.62 (SD) (3.71) -- -- (1.39) Mean Proportion 0.83 -- -- 0.17 (SD) (0.15) -- -- (0.15) Group 4 (n = 7) Mean Rate 10.37 -- -- 1.60 (SD) (3.45) -- -- (2.20) Mean Proportion 0.86 -- -- 0.14 (SD) (0.19) -- -- (0.19) Alternative Reinforcement 1 Phase RS1 RS2 RS3 Untrained Group 1 (n = 7) Mean Rate 2.86 15.91 -- 0.12 (SD) (2.15) (2.87) -- (0.33) Mean Proportion 0.15 0.85 0.01 (SD) (0.10) (0.10) (0.02) Group 2 (n = 7) Mean Rate 1.36 15.30 -- 0.08 (SD) (1.83) (1.96) -- (0.21) Mean Proportion 0.08 0.92 -- 0.00 (SD) (0.10) (0.12) -- (0.01) Group 3 (n = 5) Mean Rate 0.84 15.24 -- 0.29 (SD) (0.69) (1.73) -- (0.40) Mean Proportion 0.05 0.93 -- 0.02 (SD) (0.04) (0.02) -- (0.03) Group 4 (n = 7) Mean Rate 1.28 14.99 -- 0.30 (SD) (0.62) (3.26) -- (0.81) Mean Proportion 0.08 0.91 -- 0.02 (SD) (0.03) (0.05) -- (0.04) Alternative Reinforcement 2 Phase RS1 RS2 RS3 Untrained Group 1 (n = 7) Mean Rate 0.29 0.88 15.63 0.35 (SD) (0.37) (0.40) (3.60) (0.91) Mean Proportion 0.02 0.05 0.91 0.02 (SD) (0.02) (0.02) (0.04) (0.05) Group 2 (n = 7) Mean Rate 0.14 0.19 15.82 0.31 (SD) (0.37) (0.33) (3.41) (0.82) Mean Proportion 0.01 0.01 0.95 0.02 (SD) (0.02) (0.02) (0.08) (0.07) Group 3 (n = 5) Mean Rate 0.00 0.33 15.28 0.14 (SD) (0.00) (0.46) (2.08) (0.31) Mean Proportion 0.00 0.02 0.97 0.01 (SD) (0.00) (0.03) (0.05) (0.02) Group 4 (n = 7) Mean Rate 0.11 0.36 14.18 0.11 (SD) (0.28) (0.68) (2.36) (0.28) Mean Proportion 0.01 0.02 0.97 0.01 (SD) (0.02) (0.04) (0.06) (0.02) Resurgence Phase RS1 RS2 RS3 Untrained Group 1 (n = 7) Mean Rate 0.86 0.51 0.97 16.56 (SD) (0.61) (0.41) (0.37) (3.50) Mean Proportion 0.05 0.03 0.05 0.87 (SD) (0.03) (0.02) (0.02) (0.06) Group 2 (n = 7) Mean Rate 2.32 0.60 0.89 11.04 (SD) (1.94) (0.80) (0.78) (4.23) Mean Proportion 0.16 0.04 0.06 0.73 (SD) (0.14) (0.06) (0.06) (0.23) Group 3 (n = 5) Mean Rate 1.16 1.76 0.40 12.90 (SD) (1.19) (0.82) (0.37) (7.97) Mean Proportion 0.07 0.12 0.03 0.78 (SD) (0.05) (0.08) (0.03) (0.12) Group 4 (n = 7) Mean Rate 1.25 0.70 1.65 12.04 (SD) (0.68) (0.58) (0.99) (4.64) Mean Proportion 0.09 0.05 0.12 0.74 (SD) (0.07) (0.04) (0.07) (0.16) Note. (-) denotes no response sequence available for the phase. Mean rates include all emissions obtained in each experimental phase. The trained and untrained response sequence emissions were included in the average proportions
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||ORIGINAL ARTICLE|
|Author:||King, James E.; Hayes, Linda J.|
|Publication:||The Psychological Record|
|Date:||Sep 1, 2016|
|Previous Article:||Priority disputes in the history of psychology with special attention to the Franz-Kalischer dispute about who first combined animal training with...|
|Next Article:||The Implicit Relational Assessment Procedure (IRAP) as a measure of spider fear, avoidance, and approach.|