Presentation is loading. Please wait.

Presentation is loading. Please wait.

Technologies of Generalization

Similar presentations


Presentation on theme: "Technologies of Generalization"— Presentation transcript:

1 Technologies of Generalization
Ps620 Stimulus Control Caldwell College

2 Some Current Dimensions of Applied Behavior Analysis
Baer, Wolf & Risely (1968) Some Current Dimensions of Applied Behavior Analysis -“Generality” as one of the 7 dimensions of ABA - should be programmed, rather than lamented

3 Stokes & Baer (1977) An Implicit Technology of Generalization
-Over 50% of research reviewed relied on “train and hope” despite 1968 article’s suggestion - Article raised interest in generalization as an active process which validates the effectiveness of behavioral programming - Generalization should be pursued by Behavior Analysts - Brought realization that behavior change is most functional when its effects are widespread as well as immediate

4 Categorized 9 techniques for assessing and training generalization
Stokes & Baer (1977) Reviewed 270 studies on generalization Categorized 9 techniques for assessing and training generalization

5 Stokes & Baer (1977) Outlined 9 categories of generalization:
-train and hope -sequential modification -introduce to naturally maintaining contingencies -train sufficient exemplars -train loosely -use indiscriminable contingencies -program common stimuli -mediate generalization -train to generalize

6 Stokes & Baer (1977) 1. Train and hope
-behavior change is effected under limited conditions (persons, settings, stimuli) -assessing for generalization but not programming for it

7 Stokes & Baer (1977) 2. Sequential modification
- behavior change is effected under limited conditions (persons, settings, stimuli) -assessing for generalization THEN directly train in “generalization” settings if generalization does not occur

8 Stokes & Baer (1977) 3. Introduce to naturally maintaining contingencies -teach the target behavior to levels of performance required by naturally existing contingencies of reinforcement -hope that natural contingencies will “trap” behavior

9 Stokes & Baer (1977) 4. Train sufficient (multiple) exemplars
- Teach the student to respond to a subset of all possible stimulus and response examples and then assess generalization on untrained examples (generalization probe) Examples: Item taught, format of presentation, setting, person teaching

10 Stokes & Baer (1977) 5. Train “loosely”
Means to “randomly” vary noncritical aspects of the instructional setting within and across teaching sessions If some aspect of the materials, setting, method of presentation doesn’t matter, then change it from time to time so the learner’s response doesn’t come under control by an “irrelevant feature” this tactic often overlaps with “train sufficient exemplars” as can be seen on next slide

11 Baer’s (1999) Recommendations
Use 2 or more teachers Teach in 2 or more settings Teach from a variety of body positions (topography) Vary your tone of voice Vary your choice of words Show the stimuli from a variety of angles, using sometimes one hand and sometimes the other (topography) Have other people present sometimes and not other times Dress differently on different days Vary the reinforcers Teach sometimes in noisy settings, sometimes in quiet ones In any setting, vary the decorations, vary the furniture, and vary their locations Vary the times of day when you and everyone else teach Vary the temperature in the teaching settings Vary the smells in the teaching settings Within limits, vary the content of what’s being taught Do all of these as often and as unpredictably as possible

12 Stokes & Baer (1977) 6. Use indiscriminable contingencies
-quantity, quality, delay of delivery, & type of reinforcer should never be specific to a specific skill, setting, or teacher -same holds for schedule of reinforcement used to deliver reinforcer

13 Stokes & Baer (1977) 7. Program common stimuli (“features”)
Include aspects of the generalization setting into the instructional setting This requires analysis of “relevant features” in the generalization setting Tolerating dental procedures?

14 Stokes & Baer (1977) 8. Mediate generalization
-seems to refer to covert or overt talking to oneself

15 Stokes & Baer (1977) 8. Teach “generalizing” as a behavior
-means sometimes we need to reinforce generalization explicitly -this would be something like the opposite of “overselectivity”

16 Stokes & Baer (1977) -Article was highly acclaimed in the field of ABA
-Focus for researchers should be on functional variables responsible for generalization or lack of it CRITICAL to advancement in the field.

17 Stokes & Osnes (1989) An Operant Pursuit of Generalization
-noted the need for researchers to answer the following questions… 1. Did the behavior occur in generalized circumstances? 2. What are the functional variables which account for generalization?

18 Stokes & Osnes (1989) * Exploit Current Functional Contingencies
refined generalization promoting categories centered on the basic principles of behavior Proposed 3 categories of Generalization * Exploit Current Functional Contingencies - behaviors are naturally selected by consequences * Train Diversely- diversity in the exemplars of learning * Incorporate Functional Mediators – relationship between salient conditions of learning and stimulus control exerted over behavior by environments related to original learning

19 Stokes & Osnes (1989) * Exploit Current Functional Contingencies
Contact natural consequences Recruit natural consequences Modify maladaptive consequences Reinforce occurrences of generalization

20 Stokes & Osnes (1989) * Train Diversely
Use sufficient stimulus exemplars Use sufficient response exemplars Make antecedents less discriminable Make consequences less discriminable

21 Stokes & Osnes (1989) Incorporate Functional Mediators
Incorporate common salient stimuli Incorporate common salient social stimuli Incorporate self-mediated physical stimuli Incorporate self-mediated verbal and covert stimuli

22 Osnes & Lieblein (2003) An Explicit Technology of Generalization
25 years after Stokes & Baer (1977) 14 years after Stokes & Osnes (1989) ….have we progressed passed “train and hope” to the point where we demonstrate a generalization promoting function of behavior change procedures?

23 Osnes & Lieblein (2003) Journals from were sampled to identify articles focusing on generalization - “generalization” or “maintenance” in title or descriptor words - abstract stated that maintenance or generalization was goal or topic - condition to assess maintenance or follow-up condition present - probes for generalization - reversal design to assess durability of effect post-treatment

24 Osnes & Lieblein (2003) Journals sampled - JABA
- Behavior Modification - Journal of Positive Behavior Interventions - Behavior Analyst Today

25 Osnes & Lieblein (2003) 93 articles met the criteria
- 4 review articles - 1 discussion article - 88 research articles examined for the following: * attention on strategies for promoting generalization in Stokes & Baer (1977) or Stokes & Osnes (1989) * research methods designed to control for generalization promoting variables * explicit generalization probes * length of follow-up or maintenance conditions

26 Review Articles- all focused on social behavior
Singh, N. N., Deitz, D.E., Epstein, M. H., & Singh, J. (1991). - Reviewed 28 studies from 10 different journals - articles chosen based on: *generalization/maintenance programming – 14 articles found skill generalization and programmed for generalization across settings/people. No description of how. * follow-up assessment maintenance- less than 6 weeks (5 studies) follow-up- ranged from 2 days to 1 yr. (10 studies) (difference was unclear)

27 Singh, et al (1991) *changes in collateral behaviors resulted from programmed contingencies (2 studies reported such effects) Singh, et al. did not utilize Stokes and Baer (1977) categories of generalization but cited Stokes and Osnes (1989)

28 Review Articles Chandler, L. K., Lubeck, R. C., & Fowler, S.A. (1992). – Reviewed 51 studies from 22 journals - Analysis of 4 categories: generalization dimension generalization assessment design behavior change strategies generalization promotion strategies - Also addressed most vs. least successful generalization

29 Chandler, Lubeck & Fowler (1992)
Found 4 generalization promoting strategies used most frequently. - functional target behaviors (exploit current functional contingencies) - specification of fluency criterion (incorporate functional mediators) - indiscriminable contingencies (training diversely) - using mediation techniques (incorporate functional mediators) Described studies using Stokes and Baer (1977) and Stokes and Osnes (1989)

30 Review Articles Landrum, T.J., & Lloyd, J.W. (1992).
- Specifically examined whether generalization was explicitly addressed - 12 studies met the criteria - Authors used Stokes and Baer to guide their analysis. - Articles divided across 4 categories: 4 used “train and hope” 3 used teaching relevant behaviors 3 used sequential modification 2 used train sufficient exemplars

31 Landrum & Lloyd (1992) Maintenance assessed in 7 articles
Generalization across responses assessed in 2 articles Generalization across people assessed in 5 articles (only incidentally or anecdotally) Only 1 study assessed all 4 forms of generalization 6 studies assessed 2 forms of generalization Authors suggest generalization become a dependent variable assessed in research as Stokes and Baer did in 1968

32 Review Articles Fox, J.J., & McEvoy, M.A. (1993).
- Stated that is is necessary to not only assess but also enhance generality of interventions - “generality” vs. “generalization” used interchangeably = problematic occurrence of “generality” not sufficient must show empirical demonstration of “generalization”. - “follow-up” and “maintenance” also used interchangeably- makes determination of conditions necessary for generalization to occur difficult

33 Fox & McEvoy (1993) Reviewed research based on Stokes and Baer (1977)
Noted increases in; -generality procedures in research -diversity of tactics used -some behavior change across settings, responses and people BUT failure to replicate these effects across studies Few studies used designs which could empirically demonstrate how generality occurred

34 Discussion Article Tillman, T.C. (2000).
-Discussed generalization programming in the context of school based behavioral consultation -Framed discussion using tactics form Stokes and Baer (1977) and Stokes and Osnes (1989) -Generalization of problem-solving and intervention resulted from consultation BUT few studies actively examined generalization once examined, none could demonstrate generalization as a result of consultation. -However, Tillman discusses the possibility of creating a consultation generalization program based on Stokes and Baer and Stokes and Osnes framework for generalization

35 Research Articles Breakdown: 88 in all
- 38 used “generalization” or “maintenance” in title or descriptor words and or had a statement in the abstract - 11 included a condition to assess maintenance - 29 included a follow-up assessment of post-treatment effects - 16 specifically addressed generalization and included probes for generalization - 13 used reversal designs that allowed for assessment of durability of effects post-treatment (not maintenance per se)

36 Research Articles 38 used “generalization” or “maintenance” in title or descriptor words and or had a statement in the abstract… of those; - 30 addressed generalization (14 addressed verbal/communication behaviors, 16 addressed non-verbal behavior) - 8 addressed maintenance Bulk of research showed overt programming for generalization Exploit current functional contingencies: Craft, Alber & Heward (1998)- - taught students methods to recruit teacher praise - in generalization condition of multiple baseline design, intermittent reinforcement was used and showed improved responding that was maintained

37 Research Articles Train diversely: Halle and Holt (1991)-
- controlled for generalization by using multi-element probe design to systematically introduce various stimuli into training setting - stimulus probes revealed that paired stimuli vs. single stimulus presentation resulted in target behavior (saying please) Shore, Iwata,Lerman, & Shirley (1994)- - controlled for generalization by using diverse training and systematically varied therapist/setting/demands which resulted in varying levels of generalization of SIB (novel probes)

38 Research Articles Incorporate functional mediators:
Pierce & Schriebman (1997) - looked to increase social responding - presented a peer via multiple baseline design which increased responding of target children - also target children exhibit target behavior in novel setting w/ novel peers due to “loose training” of PRT Exploit current functional contingencies and incorporate functional mediators: Connell et al. (1993) -compared self-assessment vs. self-assessment and recruitment of teacher praise in the generalization of transition skills of children - used multiple baseline design to provide clear examination of generalization from training to classroom settings

39 Research Articles Sufficient exemplars and contacting natural consequences: Ducharme & Holborn (1997) -used multiple baseline design (ABC) -initial phase provided multiple training components to increase social skills of hearing impaired students -in second intervention phase additional peers/parents/materials/room were presented, fading teacher praise was also used -design allowed demonstration that strategies used increased in the second generalization setting (not probed post-intervention though) Multiple training exemplars: Neef et al. (1990) -compared 2 types of instruction; broad range of training exemplars to simulated vs. natural training stimuli in teaching adults w/MR to use washers and dryers -more errors in generalization were seen when broad range of exemplars was used

40 Research Articles Other articles were not designed to control for generalization but included generalization programming; Donnelly & Olczak (1990)- used DRI (exploiting current functional contingencies) to reduce cigarette pica. Used a generalization condition including other staff members. Koegel & Koegel (1990)- faded presence of trainer after criteria was met in decreasing stereotypy using self-management techniques (incorporate functional mediators) for children with autism

41 Research Articles Articles that addressed maintenance or follow-up were subdivided into research that… 1. Explicitly investigated variables resulting in maintenance 2. Assessed the presence or absence of maintenance 3. Included follow-up conditions to assess durability 4. Utilized reversal designs that allowed for examination of effect durability (expressly for maintenance assessment) Bulk of research provided maintenance assessment after treatment was withdrawn NOT designing treatment to enhance maintenance

42 Research Articles Only 4 studies actively programmed for maintenance by Exploiting current functional contingencies: Altus, Welsh, & Miller (1991)- achieved long term maintenance of task completion by members of a student housing co-op, by transferring reinforcement responsibilities from researchers to other members of the housing co-op Dozier et al. (2001) and Lerman, Iwata, & Shore (1996) used lean schedules of reinforcement and intermittent reinforcement, respectively, to maintain behavior changes achieved during treatment

43 Research Articles Incorporating functional mediators:
Bennett & Cavanaugh (1998)- studied immediate self-correction procedures vs. delayed or no self-correction procedures in multiplication tasks. Immediate self-correction was most effective in improved responding. 60% of the articles contained follow-up conditions (ranging from one session – one year) 100% of the articles that utilized a reversal design did so to demonstrate treatment effects, thus failed to show durability of effects (great for research purpose but not clinically) All used highly discriminable conditions (tokens vs. no tokens, DRA/DRL vs. none, prompts vs. no prompts) Important point when generalization (durability of effect) is desired

44 Conclusions and Recommendations
Original Question….have we progressed passed “train and hope” to the point where we demonstrate a generalization promoting function of behavior change?? Answer is mixed - researchers are most often assessing maintenance of effects BUT mostly in the hopes of demonstrating experimental control NOT in the hopes of demonstrating durability

45 Conclusions and Recommendations
To achieve both experimental control and generalization and or maintenance, researchers should take the next step..reduce highly discriminable aspects of the interventions in an additional phase of the experiment… Generalization continues to be elusive A great deal of effort is needed to achieve it…BUT it is necessary in order for Behavior Analysis to show generality of behavior change….. Without it we might as well “throw in the towel” according to the authors

46 Conclusions and Recommendations
There are examples of research studies designed to demonstrate generalization This is only a sample of journals in the field. Authors frequently referred to Stokes& Baer ( 1977) and Stokes & Osnes (1989) Generalization promotion is well entrenched in Behavior Analysis……becoming more “explicit” than “implicit”

47 Conclusions and Recommendations
BUT…we need to “raise the bar”…..NO research should be conducted without generalization promotion as a part of the research and intervention plan

48 Osnes, P.G., Lieblein, T. (2003). An explicit technology of generalization. The Behavior Analyst Today, 3,


Download ppt "Technologies of Generalization"

Similar presentations


Ads by Google