Emergence of Mands and Tacts of Novel Objects among Preschool Children

7/29/2019 Emergence of Mands and Tacts of Novel Objects among Preschool Children

1/16

The Analysis of Verbal Behavior

59

In Skinners (1957) analysis of verbal be-havior, the distinction between the mandandthe tactis a fundamental one that has far-reach-ing conceptual and practical implications. Skin-ner defined the mand as a verbal operant inwhich the response is controlled by a condi-tion of deprivation or aversive stimulation, as

a result of a history of reinforcement with aconsequence specific to that condition. Theentire class of variables that may control theform of the response in a mand relation hasalso been characterized as establishing opera-tions (EOs; Michael, 1988, 1993). In the tactrelation, by contrast, the response is controlled

by a nonverbal discriminative stimulus (SD) asa result of being reinforced with many differ-ent reinforcers or a generalized conditionedreinforcer in the presence of that stimulus

(Skinner, 1957). According to Skinners analy-

sis, any particular response form may functionas either a mand or a tact, depending on thetype of variable that evokes the response on agiven occasion. The response form water, forexample, may occur either as a mand evoked

by water deprivation, or as a tact evoked bythe sight of a glass of water. For either variable

to evoke the response, however, the responsemust have been brought under the control ofthat variable through an appropriate history ofreinforcement. In other words, the acquisitionof each operant necessarily requires a learninghistory with respect to the relevant controllingvariable. This notion of functional indepen-dence sets Skinners analysis apart from moretraditional accounts of language that assumethat the speaker acquires a word in its mean-ingful relation to a thing and then uses the word

to ask for something (p. 128). At the practicallevel, the functional independence of verbaloperants is relevant to the design of programsthat teach language, as it implies a necessity toinclude in training all variables that should ul-timately exert control over a given responseform.

Skinner (1957) noted that mand-tact inde-pendence is rarely observed among verballysophisticated speakers in the natural environ-ment. Rather, informal observation suggests

that control over the response form transferseasily between EO and SD, such that followingthe acquisition of a tact, even young childrenwill readily emit the same response form as amand, and vice versa. Skinner provided at least

Emergence of Mands and Tacts of Novel

Objects among Preschool Children

This article is based on a thesis completed by thefirst author, and supervised by the second and thirdauthors, that was submitted to The Graduate Col-lege at Western Michigan University in partial ful-fillment of the requirements for the Master of Artsdegree in Psychology. We thank the directors andstaff of the child care centers in which the studywas conducted for their helpful support, and we alsothank the many undergraduate students who assistedwith data collection; in particular, Veronica Hill, and

Meghann Mitchell.Address correspondence to James E. Carr, De-

partment of Psychology, Western Michigan Univer-sity, 1903 W. Michigan Ave., Kalamazoo, MI,49008-5439; e-mail: [email protected].

2005, 21, 5974

Anna Ingeborg Petursdottir, James E. Carr, and Jack Michael

Western Michigan UniversityAccording to Skinners (1957) analysis of verbal behavior, the mand and the tact are functionally indepen-dent verbal operants, each of which is acquired through a unique history of reinforcement. The presentstudy attempted to replicate the findings of Lamarre and Holland (1985), who empirically demonstratedfunctional independence of mands and tacts in typically developing preschool children. Five children

participated. All were initially trained to complete two 4-piece assembly tasks. Four children were thentrained to tact the four pieces that comprised one of the assembly tasks, and to mand for the four pieces thatcomprised the other task, using arbitrary vocal response forms. The remaining child received tact trainingonly, and only on one task. The effects of training on the untrained operant were evaluated in a multiple-

probe design across tasks. Following mand training, 4 out of 4 children reliably emitted tacts under testingconditions, while the effects of tact training differed across participants. The results differ from those ofLamarre and Holland, but are not necessarily surprising from the point of view of either Skinners analysisor more recent behavioral accounts of language. Future research should attempt to identify variables thataffect transfer between mand and tact relations.

Key Words: mand, tact, functional independence, verbal behavior, children.


2/16

60 ANNA I. PETURSDOTTIRet al.

four reasons why such transfer might occur.First, tact emergence may be facilitated by theacquisition of a mand in the presence of themanded stimulus. Second, similarity between

the stimulus that evokes a tact and the stimu-lus that reinforces a mand may facilitate trans-fer from one operant to another. Third, trans-fer may occur if caregivers reinforce one oper-ant as if it were the other; and fourth, childrenmay early in life acquire generalized verbalskills that permit efficient acquisition of bothmands and tacts. Given the number of condi-tions that can possibly give rise to transfer be-tween EO and SD control, mand-tact indepen-dence may rarely be observed except in veryyoung children or in individuals who are notsufficiently responsive to the natural environ-ment to allow for typical language acquisition.

Skinner (1957) did not consider childhoodlanguage acquisition in detail, and thus did not

provide any suggestions as to how late in a typi-cally developing childs life it might be pos-sible to observe mand-tact independence. Con-temporary behavioral accounts of languagehave suggested specific higher-order verbalskills that, once acquired, override the func-tional independence of mands and tacts. Rela-

tional frame theory (Hayes, Barnes-Holmes, &Roche, 2001) suggests that once a child hasacquired the ability to derive arbitrary stimu-lus relations, untrained mands and tacts mayoccur for all stimuli that participate in relationalframes with other stimuli that may be mandedor tacted in similar contexts (Barnes-Holmes,Barnes-Holmes, & Cullinan, 2000). By con-trast, an account proposed by Horne and Lowe(1996) suggests that the critical skill is nam-ing, which is hypothesized to be a higher-or-

der verbal operant consisting of interlockingtact relations, echoic relations, and listener

behavior. According to this account, whenmands enter into name relations, they becomeinstances ofname-manding. Tacts and mandsmay exist in isolation when not accompanied

by name relations, but in typical language ac-quisition, this should rarely be the case oncenaming skills have been established. Neitherrelational frame theory nor the naming accounthas specified a timeline, in terms of chrono-

logical age, for when tact-mand independencemight be overcome. However, typically devel-oping children of preschool age have beenshown to perform in accordance with the hy-

pothesized higher-order skills; that is, arbi-

trarily applicable relational responding orstimulus equivalence (e.g., Goyos, 2000;Saunders, Drake, & Spradlin, 1999; Sidman,Willson-Morris, & Kirk, 1986) and naming

(Horne, Lowe, & Randle, 2004; Lowe, Horne,Harris, & Randle, 2002).Nevertheless, Lamarre and Holland (1985)

successfully demonstrated functional indepen-dence of mands and tacts of typically develop-ing 3- to 5-year-old children. Nine childrenwere taught to emit the responses on the leftand on the right as tacts (tacting the relativelocation of objects) and mands (manding forthe experimenter to place an object either tothe left or to the right of another object). Theuntrained operant was then tested under extinc-tion. In no case did the training of one operantreliably result in the emergence of the otherunder testing conditions; some children neverresponded correctly in the untrained functionand others did so only on a few occasions.Hence, all participants required direct trainingin both functions. Furthermore, when the ex-

perimenters reversed the traditional conceptsof left and right, and re-trained either tacts ormands under the new contingencies, the ma-

jority of the children did not alter their re-

sponses when tested in the untrained function.This was the case even though mand trainingand testing were accomplished in the presenceof the SD for the tact; that is, the left and rightlocations were always visible to the partici-

pants.Lamarre and Hollands (1985) results might

suggest that transfer between mand and tactcontingencies does not occur with the ease sug-gested by Skinner (1957), Barnes-Holmes etal. (2000), or Horne and Lowe (1996), and that

preschool-age children do not acquire both re-lations unless the response is directly reinforcedin the presence of both the relevant EO and therelevant SD. However, several applied studiesthat have investigated the functional indepen-dence of mands and tacts of children and adultswith developmental disabilities (Hall &Sundberg, 1987; Nuzzolo-Gomez & Greer,2004; Sigafoos, Doss, & Reichle, 1989;Sigafoos, Reichle, Doss, Hall, & Pettitt, 1990;Twyman, 1996) have demonstrated a greater

degree of transfer than was observed byLamarre and Holland. Although functional in-dependence was demonstrated in all of thesestudies, some or all of the participants in eachstudy acquired some mands or tacts without


3/16

61MAND AND TACT EMERGENCE

direct training. In two studies, a brief historyof training in both functions for additional tar-get responses was followed by reliable mand-tact transfer (Hall & Sundberg; Nuzzolo-

Gomez & Greer). Interestingly, the participantsin all of those studies either had minimal ver-bal repertoires (Hall & Sundberg; Sigafoos etal., 1989; 1990), or substantial language de-lays (Nuzzolo-Gomez & Greer; Twyman), andat least a similar, if not a greater, degree ofmand-tact transfer in preschoolers without suchimpairments might be expected.

It is possible that some aspect of the prepa-ration or procedures used in Lamarre andHollands (1985) study was responsible for theabsence of untrained operants under testingconditions. Some evidence that this may be thecase has been presented by Egan and Barnes-Holmes (2004). These researchers initially rep-licated Lamarre and Hollands results withyoung children diagnosed with autism, butfound that modifications to the testing proce-dures, intended to provide effective contextualcues for responding, resulted in the emergenceof untrained operants. It is thus possible thatthe children in Lamarre and Hollands studywere simply unable to respond effectively to

the testing questions. Another aspect of theprocedures that may have prevented transferconcerns the presence of EO control in mandtraining and testing conditions. For the re-sponses on the left and on the right to func-tion as mands, sometimes an EO would haveto be in effect that increased the reinforcingvalue of seeing an object placed on the left,and at other times an EO would have to be ineffect that increased the reinforcing value ofseeing an object placed on the right. The re-

searchers did not manipulate these EOs, and itis unclear exactly what they consisted of orwhether they were at all present. If during mandtesting, no EO was present that increased thereinforcing value of either location, the re-sponses on the left or on the right wouldnot be expected to occur, even though they had

been acquired as tacts. Similarly, if duringmand training, no EO was present to controlthe form of the response, it is possible that theresponses on the left and on the right were

established primarily as intraverbal responsesto the experimenters question, Where do youwant me to put the [object]?, and thereforedid not occur in response to the questionWhere is the [object]? used on tact probes.

The present study was an attempt to system-atically replicate Lamarre and Hollands (1985)results, while manipulating the presence of EOsduring mand training and testing. Young chil-

dren were trained to tact and mand for objectsthat comprised assembly tasks. The childrenreceived rewards upon completing the tasks.The mand condition utilized an interrupted-chain procedure similar to that used by Halland Sundberg (1987) and Carroll and Hesse(1987). This procedure involved instructing thechildren to complete assembly tasks in the ab-sence of items needed to complete them. Thesight of an incomplete assembly task was in-tended to function as an EO that rendered themissing piece effective as reinforcement. Thisstudy differed from Lamarre and Hollandsstudy in several ways besides the presence ofEO manipulation. First, the target responseswere mands and tacts of objects rather thanlocations or other abstract stimulus properties.Second, in order to prevent the participantsfrom acquiring the target mands and tacts out-side of the experimental situation, the responseforms consisted of nonsense words rather thanconventional English response forms. Third,training and testing of each operant were con-

ducted in the absence of the controlling vari-able for the other operant; in other words, theSDs for tacts were never present on mand train-ing or testing trials, and the EOs for mands werenever present on tact training or testing trials.Fourth, the children in this study were youngerthan those in the Lamarre and Holland study.The last three features were incorporated spe-cifically in order to increase the probability ofdemonstrating functional independence.

METHOD

Participants and Setting

Five children participated in the study:Tristan, Emma, Noah, Olivia, and Mackenzie.Their ages ranged from 2 yrs., 6 mos. to 3 yrs.5 mos. at the time they entered the study, andfrom 2 yrs., 9 mos. to 3 yrs. 8 mos. at the endof the study (see Table 1 for participants agesat the beginning of each test condition; note

that age in baseline in some cases differed fromage at study onset because of an interveningpretraining phase). None of them had anyknown developmental delays, based on parentreport. The children were recruited from three


4/16


local child-care centers that gave permissionfor the study to be conducted in their facilities.Parent permission and child assent were ob-

tained prior to each childs entry into the study.Experimental sessions were conducted eitherin the childrens classrooms or open areas out-side of their classrooms. A number of childrenwere typically present and engaged in regularactivities while sessions were conducted. Dur-ing all sessions, a child and an experimenterwere seated across from or next to each otherat a child-sized table. Each session lasted ap-

proximately 10-15 min, and each child wasscheduled to attend 1 or 2 sessions per day, 3to 5 days a week. A camcorder used to recordtesting sessions was located on a tripod behindthe experimenter.

Materials and Programmed Consequences

Two assembly tasks were constructed for thepurpose of the study. The cube consisted of fourfoam pieces, each made of 6 or 7 foam blocksthat had been glued together to form a uniquethree-dimensional shape. When the four pieceswere assembled correctly, they formed the

shape of a cube. The puzzle contained fourwooden puzzle pieces that differed from oneanother in shape and the location of coloreddots on a solid-color surface. One assemblytask was used in each session of the experi-

ment. In addition, a variety of common toys(e.g., cars, dolls, animals) were used duringscreening and testing sessions.

At the beginning of each session, the childreceived a sticker sheet on which to collectstickers obtained as consequences for correctresponding during the session. A variety ofstickers were available throughout the experi-ment, and prior to each session, the child se-lected the type of stickers that he or she wishedto receive in that session. During all sessions,stickers, along with praise, were delivered con-tingent on specific correct responses. At the endof each session, the child was allowed to keepthe sticker sheet with the stickers that he orshe had earned.

Data Collection

Dependent variable. The primary dependentmeasure was untrained verbal operants (mandsor tacts) emitted under testing conditions. A tactwas defined as a response made when the ex-

perimenter held up a piece belonging to one ofthe assembly tasks, and asked, What is this?A mand was defined as a response made when

one of the pieces needed to complete an as-sembly task was out of sight, and the experi-menter asked What do you need? The targetresponse forms were determined by one- ortwo-syllable names that were assigned to each

Table 1Participants chronological ages (in months) at the beginning of each testing condition,

and response forms used for each participant.

Age in MonthsBaseline Post-training Cube Puzzleonset 1 2 3 4 words words

Tristan 41 43 43 n/a n/a chey, noo boosha, dososai, wa heeny, middy

Emma 37 42 43 44 n/a meep, wak bindow, lacketsoof, trog meecot, nover

Noah 40 41 43 n/a n/a doob, gop boosha, heenykig, neek voggy, noker

Olivia 37 38 39 40 41 meep, wak bindow, lacketsoof, trog meecot, nover

Mackenzie 31 33 n/a n/a n/a meep, wak bindow, lacketsoof, trog meecot, nover


5/16


of the eight pieces comprising the two assem-bly tasks prior to baseline testing for each child.One-syllable names were assigned to cube

pieces and two-syllable names to puzzle pieces.

The specific names were selected and/or modi-fied based on each childs echoic repertoire,with the restriction that they could not be listedas words inMerriam-Websters Collegiate Dic-tionary (Mish et al., 1993). Table 1 lists thetarget responses for each child.

Scoring. On each trial, during both testingand training sessions, the experimenter re-corded either a correct or an incorrect responseon a data sheet. A response was scored as cor-rect if it contained the target response form anddid not contain any of the other response formsdefined for that child. A response was scoredas incorrect if the target response form was notemitted within 20 s of the experimenters ini-tiation of a testing trial, or within 5 s of theinitiation of a training trial; or if another re-sponse form defined for the child was emittedeither before or within 5 s of the target responseform.

Interobserver agreement. A second observerindependently recorded data for 91% of all test-ing sessions, either while present during the

session or subsequently from videotape. Oneach trial, an agreement was scored if the ex-

perimenter and the second observer both scoreda response as correct or incorrect; otherwise, adisagreement was scored. Point-by-pointagreement was calculated for each session bydividing the number of agreements by the sumof agreements and disagreements, and multi-

plying by 100%. Agreement ranged from87.5% to 100% for individual testing sessionsand averaged 97.2% for Tristan, 98.7% for

Emma, 99.3% for Noah, 99.6% for Olivia, and98.9% for Mackenzie.

During training, a second observer waspresent and independently recorded data for45% of all sessions. Point-by-point agreementwas calculated for each session in the samemanner described for testing sessions. Agree-ment ranged from 87.5% to 100% for indi-vidual training sessions and averaged 99.5%for Emma, 99.2% for Olivia, and 100% for theother three children.

Independent variable integrity. The indepen-dent variable was the training of a verbal oper-ant (mand or tact). A trained observer recordedexperimenter behavior for 41% of all trainingsessions. A training error was scored if during

the training of one operant, the experimenterdelivered consequences appropriate only fortraining of the other operant (e.g., delivering

praise following a correct response on a mand

trial). An error was also scored if the experi-menter failed to deliver consequences or errorcorrection procedures. A trial was scored ascorrectly implemented if no errors were madeon that trial. All scored trials (100%) wereimplemented correctly during tact training and98.6% were implemented correctly duringmand training.

Procedures

Screening. Prior to the experiment, prospec-tive participants were screened for two prereq-uisite skills: (a) Tacting familiar toys (e.g., atoy car) in response to the question What isthis? and (b) manding for familiar toys thatwere absent and needed to complete play rou-tines (e.g., a baby bottle needed to feed a babydoll) in response to the question What do youneed? The purpose of screening was to verifythat the instructions used during testing weresufficient to evoke mands and tacts already inthe repertoire. To qualify for participation in

the study, a child had to respond correctly oneither 2 out of 2 or 3 out of 4 trials of eachtype. No children were excluded from the studyon the basis of not meeting this criterion.

Pretraining. In the pretraining phase, thechildren were trained to assemble the cube andthe puzzle. Backwards chaining was used toteach the cube and total-task presentation forthe puzzle. The experimenter began each trial

by delivering the instruction Put these to-gether (or Finish putting these together,

prior to the final step of training on the cube).Initially, the experimenter assisted the childwith the completion of the task, followingwhich the child received praise and a sticker.Over subsequent trials, the experimenters as-sistance was gradually withdrawn until thechild was performing the task independently.Training continued until the child completedthe task independently on 3 out of 3 trials ontwo consecutive days.

Tact training. In tact training, the children

were trained to tact each of the four pieces thatcomprised one of the assembly tasks. Trainingwas conducted in a discrete-trial format. Oneach trial, the experimenter held up one pieceat a time and asked, What is this? If the child


6/16


responded correctly, the experimenter deliveredpraise (e.g., Very good!) and a sticker. If thechild made an incorrect response or made noresponse within 5 s, the experimenter vocally

prompted a correct response (e.g., a wak),and then repeated the trial.Training proceeded through the following

steps. Trials were first conducted with piece 1until the child responded correctly, without

prompts, for three trials in a row. Piece 2 wasthen introduced in the same fashion, follow-ing which presentations of pieces 1 and 2 werealternated such that every 4-trial block con-tained two presentations of each piece. When100% correct responding was achieved on one

block of trials, piece 3 was introduced, and itspresentations alternated with piece 2 in thesame manner. After 100% correct respondingoccurred for one block of trials, piece 4 wasintroduced and alternated first with piece 3 andthen with piece 1, at which time presentationsof each piece had been alternated with twoother pieces. In the final stage of training, pre-sentations of all four pieces were alternated in8-trial blocks, in which each piece was pre-sented twice, with presentation order varyingacross blocks. Training was completed when

the child responded correctly on at least 7 outof 8 trials in three consecutive 8-trial blocksthat were conducted on at least two separatedays. For Olivia, who received tact training onthe cube twice, the initial steps of training wereomitted the second time, and, thus, training

began with alternating presentations of all fourpieces.

Mand training. In mand training, the childrenwere trained to mand for each of the four piecesthat comprised one of the assembly tasks. Train-

ing was conducted in a discrete-trial format.Prior to each trial, the child was presented with3 of the 4 objects needed to complete the as-sembly task and given the instruction Put thesetogether. The fourth object was kept out of thechilds sight but within the experimenters reach.When the child had attempted to complete thetask, the experimenter asked, What do youneed? (if a correct mand occurred prior to thequestion, the question was omitted). If the childresponded correctly with the name of the miss-

ing piece, the experimenter immediately deliv-ered that piece without simultaneously provid-ing any other consequences such as praise orsmiles. The receipt of the missing piece enabledthe child to successfully complete the task. The

experimenter then praised task completion (e.g.,Good job putting them all together!) and gavethe child a sticker.

If the child did not respond to the question

What do you need? within 5 s, the experi-menter vocally prompted a correct response(e.g., the wak), following which the trial wasrepeated, beginning with the instruction tocomplete the assembly task. If an incorrect re-sponse occurred, such that instead of respond-ing with the name of the missing piece the childresponded with the name of one of the piecesthat he or she already had on the table, the ex-

perimenter picked up that piece and gave itback to the child, following which a correctresponse was prompted and the trial repeated.The purpose of this error correction procedurewas to create a history in which the sight of theincomplete task functioned as an EO that in-creased the reinforcing value of the missing

piece and did not function as an SD correlatedwith the availability of reinforcement. That is,a mand for a particular piece (e.g., I need thewak) always resulted in the delivery of that

piece and was thus not correlated with the avail-ability of the consequence, but the delivery ofthe piece was assumed to be reinforcing only

if it happened to be the missing piece neededto complete the task.

The training steps and acquisition criteriawere the same as those used for tact train-ing. For Emma and Olivia, who receivedmand training on the puzzle and the cube,respectively, after already receiving tacttraining on those tasks, the initial steps oftraining were omitted in this additional mandtraining phase, such that training began withalternating presentations of trials targeting

all four mands.Testing. Testing sessions were conducted in

baseline and following each training phase.Each testing session contained eight test trials

presented in a variable order: one mand trialand one tact trial for each of the four piecesthat comprised one of the assembly tasks. Testtrials were identical to training trials with theexception that they were conducted under ex-tinction; that is, no consequences were pro-vided for either correct or incorrect responses.

Instead, up to 20 s were allowed for a responseto occur and the task materials then removed.Modified testing trials for mands were used

for Olivia only. On modified mand trials, non-specific mands such as I need another piece


7/16


in response to the initial question What do youneed? were followed by prompts to mand fora specific piece (Which one?).

Test trials were interspersed with two types

of reinforced trials: (a) tact trials in which fa-miliar items (e.g., a toy car or a crayon) werepresented along with the question What isthis?, and (b) successful task completion tri-als, in which the child was presented with allfour pieces of the assembly task, along with aninstruction to complete the task. In both typesof trials, the child received praise and a stickerfollowing a correct or a prompted response.Each test trial was followed either by one fa-miliar tact trial or one successful task comple-tion trial. The purpose of these trials was (a) tomake reinforcement available during testingsessions, and (b) to prevent extinction of taskassembly behavior by providing opportunitiesto complete the task successfully.

Before the first testing session in each phasefollowing baseline, a booster pretraining ses-sion was conducted for each assembly task inorder to ensure that the child could still com-

plete the task independently. If the child hadreceived mand training on one of the tasks im-mediately prior to the testing phase, the

booster training session for that task was omit-ted.

Experimental design. The effect of trainingone operant on the emission of the untrainedoperant was evaluated in a multiple-probe de-sign across the two assembly tasks. Four chil-dren (Tristan, Emma, Noah, and Olivia) re-ceived tact training on one task and mand train-ing on the other task, with the order of tasksand training phases counterbalanced across

participants. The fifth child, Mackenzie, com-

pleted tact training on the puzzle, but discon-tinued daily attendance at the child-care cen-ter before mand training could be completed.

If the untrained operant did not emerge atcriterion level following training, that operantwas trained separately following initial train-ing on both tasks. Criterion was met if withinsix testing sessions, at least one session oc-curred in which the child responded correctlyon 4 out of 4 trials in the untrained function, orat least two sessions occurred in which three

correct responses were made. If this criterionwas not achieved within six testing sessions, anew training condition commenced, except thatadditional modified testing sessions were con-ducted for Olivia.

RESULTS

Tristan (Figure 1) and Emma (Figure 2) re-ceived mand training first, followed by tact

training. In the first testing session followingmand training on the puzzle, Tristan respondedcorrectly on 3 out of 4 tact trials, and in thetwo subsequent sessions, he responded cor-rectly on all four. Following tact training onthe cube, Tristan never responded correctly onall four mand trials in a single session. His er-rors, however, were not limited to mands forspecific pieces; he emitted each of the fourmands correctly in at least one session. Fol-lowing this second training phase, performanceon mands and tacts acquired during the firsttraining phase had deteriorated.

Emma responded correctly on 3 out of 4 tacttrials following mand training on the cube, andin the second session, she responded correctlyon all four trials. Even though Emma had pre-viously completed mand training, she neveremitted all four mands in a single testing ses-sion; however, she emitted each of the fourmands correctly in at least two sessions. Fol-lowing tact training on the puzzle, Emmascorrect mands ranged from 0 to 3 in each ses-

sion. As with Tristan, her errors on mand trialswere not limited to mands for specific pieces;she emitted each of the four mands correctlyin at least one session. Because Emmas per-formance on mand trials did not meet the ac-quisition criterion, she subsequently receivedmand training on the puzzle. Following mandtraining, both mands and tacts occurred reli-ably. As with Tristan, Emmas performance intesting sessions for mands and tacts acquiredduring mand training deteriorated following the

completion of tact training.Noah (Figure 3) and Olivia (Figure 4) re-

ceived tact training first, followed by mandtraining. In the first session following tact train-ing on the puzzle, Noah did not make any cor-rect responses on mand trials. However, per-formance increased across subsequent sessionsand in the fifth session he responded correctlyon all four mand trials. Following mand train-ing on the cube, Noahs performance on bothtacts and mands was variable and less than

perfect, but on the sixth probe he respondedcorrectly on all four tact trials. Mands and tactsfrom the first training phase were not main-tained following the second phase.

Olivia did not respond correctly on any mand


8/16


Figure 1. The effects of tact and mand training on tact (filled circles) and mand (open squares) acquisition for Tristan.

trials following tact training on the cube; in-stead, she responded with a nonspecific mand(I need another piece.) on all trials. Modi-fied mand trials, which included a prompt tomand for a specific piece, did not result in anycorrect mands. In the first session followingmand training, Olivia also failed to respondcorrectly on all four tact trials. However, her

performance increased across subsequent ses-sions and in the third session she respondedcorrectly on all four tact trials. The tacts thatOlivia acquired during tact training on the cubewere not maintained following mand trainingon the puzzle. Therefore, tacts on the cube were

retrained prior to introducing mand training onthis task. Following the second round of tacttraining, Olivia responded correctly on all fourmand trials in the first testing session, but cor-rect mands were rarely observed in subsequentsessions. Following mand training on the cube,correct mands and tacts were reliably observed.

Mackenzies data are depicted in Figure 5.

Mackenzie received tact training on the cube,following which she never emitted any correctmands. As with Olivia, her incorrect responsesfrequently, but not always, consisted of non-specific mands such as I need another blockor I need another one.


9/16


Figure 2. The effects of tact and mand training on tact (filled circles) and mand (open squares) acquisition for Emma.

As shown in Table 2, the mean number ofuntrained tacts following mand training wasgreater than the number of untrained mandsfollowing tact training for all four children whoreceived both types of training. Table 2 alsoshows each childs performance on the un-trained operant in the first testing session fol-lowing each training phase. Although no re-

sponses were reinforced on testing trials, theincreasing trend across testing sessions for Noah(following tact training on the puzzle) andOlivia (following mand training on the puzzle)indicates a possibility that some learning nev-ertheless occurred during testing sessions. Data

on the first session following each training phasemay therefore represent the effects of trainingalone, without any possible confounding effectsof testing. For 3 out of 4 children who receivedmand training, three correct tacts were observedin the first testing session following training,while none were observed for Olivia. Tristanalso responded correctly on three mand trials

in the first session following tact training, butthe other four children emitted either one cor-rect mand or none.

Figure 6 shows the number of training trialsrequired to reach criterion in each training

phase for each child. For 3 of the 4 children


10/16


Figure 3. The effects of tact and mand training on tact (filled circles) and mand (open squares) acquisition for Noah.

who completed both mand and tact training,the first training phase took longer to completethan the second phase. Also, mand training tooklonger to complete than tact training for 3 outof 4 children.

DISCUSSION

In the present study, mand training resultedin reliable tact acquisition for 4 out of 4 chil-dren, but tact emergence was delayed for oneof them. Tact training, on the other hand, had aless consistent effect on mand acquisition. Spe-

cifically, two children did not emit any mandsfollowing tact training, one began to emit cor-rect mands only after repeated exposure to test-ing sessions, and while the remaining two chil-dren responded correctly on some mand trialsin most testing sessions, they never emitted allfour mands correctly in a single session.

These results differ from Lamarre and

Hollands (1985) study in which preschoolchildren completely failed to exhibit appropri-ate tact repertoires following mand training, aswell as mand repertoires following tact train-ing. The present study differed from that of


11/16


Figure 4. The effects of tact and mand training on tact (filled circles) and mand (open squares) acquisition for Olivia.

Lamarre and Holland in two ways that mayhave contributed to the different outcomes.First, Lamarre and Holland taught mands andtacts of relative locations of objects, which arearguably more abstract than the mands and tactsof discrete objects taught in this study. Youngchildren may have more history with respectto manding and tacting objects than they do

with manding and tacting relative locations,and even though the children in the presentstudy were substantially younger than those inLamarre and Hollands study, it is conceivablethat such a history facilitated transfer between

operants. Second, the present study utilized aninterrupted-chain procedure in an attempt tocontrive EOs during mand training and test-ing. In the Lamarre and Holland study, by con-trast, EOs were not under the control of theresearchers and may or may not have been

present during mand training and testing. Asdiscussed earlier, if it was the case that EOs

were absent, these conditions may not havebeen true mand conditions, which implies thatno transfer of control from EO to SD or SD toEO was possible. Given the discrepant results,it appears that further attempts to replicate


12/16


Lamarre and Hollands findings are warranted.Future research should continue to incorporatevariables that promote EO control over allegedmand responses and attempt to demonstratefunctional independence across a variety ofstimuli or events to be manded or tacted.

A second aspect of the present results thatwarrants discussion is that mand training pro-duced a more consistent effect on the tact rep-ertoire than did tact training on the mand rep-

ertoire. One potential explanation is that self-echoic responding during mand training facili-tated mand-to-tact, but not tact-to-mand, trans-fer. Anecdotally, all of the children were ob-served to occasionally engage in echoic behav-

ior following reinforcement during mand train-ing. After successfully manding for and receiv-ing a missing piece, the children would echotheir own mands in the presence of the piecewhile completing the assembly task. Whilethese self-echoic responses were observed onlysporadically (formal data were not collected ontheir occurrence, but experimenters notes in-dicate that they were observed on more thanone occasion with each child), it is possible

that they occurred more frequently at a covertlevel. If that was the case, then self-echoic re-sponding following delivery of the reinforc-ing piece may have served to establish discrimi-native control of that piece over the response

Figure 5. The effects of tact training on tact (filled circles) and mand (open squares) acquisition for Mackenzie.


13/16


Figure 6. Trials to criterion in each training phase for each participant.

Table 2Mean Number of Untrained Responses Following the Initial Mand and Tact Training Phases,

and Number of Untrained Responses in First Testing Session Following Training.

1For untrained responses following the first training phase, the mean number of responses is based on sessions con-ducted prior to the second training phase.

form. During tact training, no analogous op-portunity was present for the incomplete taskto acquire EO control over the response form,which may explain why mand training wasmore successful than tact training in produc-ing both operants.

It should be noted that Skinner (1957) ap-pears to have suggested an asymmetrical rela-tionship between tact-to-mand transfer andmand-to-tact transfer, as he stated that perhapsall mands which are reinforced by the pro-

duction of objects or other states of affairs maybe interpreted as manding the behavior of thelistener and tacting the object or state of af-fairs to be produced (1957, p. 189). As noted

by Horne and Lowe (1996), this statement isconsistent with their notion of name-manding.It appears to be the case that any time a mand

is reinforced, a stimulus is presented that mayacquire discriminative control over a tact re-sponse identical to the mand, which perhapsoccurs through self-echoic responding if thestimulus is present only following, but not priorto, the occurrence of the mand. By contrast,when a tact is reinforced, there is not necessar-ily an EO present to acquire adventitious con-trol over the response and establish it as a mand.The form of a response in a mand relation maysometimes be exclusively under EO control,

such as when an infants crying takes differentforms when the infant is hungry and when theinfant is wet. However, both Skinner (1957)and Horne and Lowe (1996) appear to suggestthat if the form of the mand is identical to thatwhich the verbal community recognizes as atact of the reinforcing stimulus, then the mand

Untrained in first

Mean untrained1 sessionMands Tacts Mands Tacts

Tristan 2.50 3.67 3 3Emma 1.67 3.33 1 3

Noah 2.20 2.67 0 3Olivia 0.00 2.25 0 0Mackenzie 0.00 n/a 0 n/a


14/16


is unlikely to occur unless the tact is also in therepertoire. Hence, the training of a mand maynecessarily involve implicit tact training,whereas tact training involves no such implicit

mand training. The present results appearmostly consistent with that notion. A possibleexception, however, is Olivias performance inthe first testing session following mand train-ing. The fact that she did not respond correctlyon any tact trials in this session may indicatethat mand training alone did not suffice to es-tablish tacts. The emergence of tacts in subse-quent testing sessions remains unexplained, butin line with the previous analysis, it could bethat the presence of tact trials during testingsessions prompted increased self-echoic re-sponding, or tact rehearsal, on mand trials.

In addition to the Lamarre and Holland(1985) study, two other studies (Twyman, 1996;

Nuzzolo-Gomez & Greer, 2004) have demon-strated an absence of mand-to-tact transfer inchildren who had already acquired a numberof both mands and tacts for various objects. In

both studies, however, transfer of control mayhave been complicated by the fact that mandsand tacts with novel response forms were su-

perimposed on existing mands and tacts with

less complex response forms. Further, partici-pants in both studies had impaired verbal rep-ertoires. An area for future research might in-volve exploring further the extent to which, andunder what conditions, mands occur in the ab-sence of tacts.

While the effects of mand training were simi-lar across children, the effects of tact trainingdiffered from one child to another. The vari-ables responsible for these differences are un-known; however, participant characteristics

that may have influenced the results includechronological age and the order of training

phases within the study. The two children whodid not emit any mands following tact train-ing, Olivia and Mackenzie, were also theyoungest two participants at the time that tacttraining was completed and testing began.Olivia was 3 yrs., 2 mos. old and Mackenziewas 2 yrs., 9 mos. old, whereas the otherchildrens ages at the completion of tact train-ing ranged from 3 yrs., 5 mos. to 3 yrs., 7 mos.

It seems possible that the older children hadmore advanced verbal skills that resulted inmore effective tact-to-mand transfer. It should

be noted that Olivia was also the youngest childtested for tact emergence at the completion of

mand training, and the only one for whom tactacquisition was not demonstrated immediately.Mackenzies participation in the study wasunfortunately discontinued before it was pos-

sible to observe the effects of mand trainingon her tact repertoire.Egan and Barnes-Holmes (2004) reported

that mand-tact independence observed in ini-tial testing sessions was overcome when ante-cedent instructions were modified, but this wasnot the case for Olivia in the present study.Olivia did not respond correctly on any mandtrials following initial tact training, even whenmand probes were modified in an attempt toevoke the response forms that had been trainedas tacts. It is possible that the modificationimplemented in the present study was simplynot effective, and that a yet more explicit in-struction (e.g. What is the name of the pieceyou need?) would have evoked correct mands.However, it should be noted the tasks and in-structions used in the present study differedfrom those used by Egan and Barnes-Holmesin several ways; including that in the presentstudy, the stimuli to be manded were not vis-ible to the children.

With respect to the three children who did

acquire mands as a result of tact training,Tristan and Emma showed a similar pattern ofresponding, although Tristan responded cor-rectly on a greater percentage of mand trialsthan Emma did. Noah exhibited a different

pattern, as mands emerged only in the secondtesting session following tact training. On thefirst testing trial on which a correct mand oc-curred (in session 6), Noah was observed to

point to and tact the three pieces on the table,following which he successfully manded for

the fourth piece when the experimenter asked,What do you need? This behavior may haverepresented a strategy that enabled mand ac-quisition, not as a direct result of training, butas a result of exposure to the testing conditions.Emma and Tristan, by contrast, emitted cor-rect mands already in the first session follow-ing tact training (although Emma respondedcorrectly on only one trial). Emma and Tristandiffered from Noah in that by the time theyreceived tact training, they had already received

mand training on another assembly task. It ispossible that this prior exposure to mand train-ing in some way facilitated mand acquisitionduring tact training. Consistent with that hy-

pothesis, when Olivia received a second round


15/16


of tact training on the cube, after being exposedto mand training on the puzzle, she initiallyresponded correctly on all four mand trials,although the mands did not maintain over sub-

sequent testing sessions. These results appearconsistent with studies that have demonstratedtransfer following a history of training in bothfunctions under similar conditions (Hall &Sundberg, 1987; Nuzzolo-Gomez & Greer,2004). Further, the results might be conceptu-alized as indicative of the type of history pro-

posed by Barnes-Holmes et al. (2000) to resultin the emergence of untrained mands and tacts.Future investigations of mand-tact indepen-dence might focus on younger children andsystematically explore participant variables thatmay be correlated with the emergence of un-trained mands and/or tacts. In particular, it may

be important to investigate whether any pre-requisite verbal skills or prerequisite traininghistories are required for transfer between op-erants. The identification of such skills or train-ing histories would have implications for lan-guage assessment and intervention selection.

Some limitations of the present study maybe noted. The first concerns the presumed ab-sence of multiple stimulus control in the tact

and mand conditions. Training and testing con-ditions in this study were specifically designedsuch that the presumed controlling variablesfor each operant (EOs for mands and SDs fortacts) were present only during training andtesting of that operant. However, some stimu-lus generalization may have occurred acrossconditions due to possible physical similarity

between the shape of each piece and the shapeof an incomplete assembly task without that

piece. For example, the shape of each puzzle

piece may have been similar in shape to thegap in the puzzle when that piece was miss-ing. It is conceivable that this resemblancecaused the sight of the incomplete puzzle on amand testing trial to evoke a response previ-ously trained as a tact in the presence of themissing piece. On a tact testing trial, similarly,the sight of a puzzle piece might have evokedthe response that was already under EO con-trol of the gap in the puzzle. However, if it wasthe case that multiple control was introduced

this way, it does not explain why mand train-ing had a greater effect on the untrained oper-ant than did tact training, and also does notexplain why transfer between operants wasobserved in this study to a greater extent than

in Lamarre and Hollands (1985) study inwhich the presence of the SD for the tact in themand condition was more salient and did notrequire stimulus generalization.

A second potential limitation is that the studywas conducted in a relatively uncontrolled en-vironment that contained a number of poten-tially disruptive stimuli. Noise level variedunsystematically across sessions, and sessionswere intermittently interrupted by other chil-dren. Variations in the level of disruption may

be partially responsible for the variability inresponding across testing sessions that wasobserved for all five children. Therefore, al-though the first testing session following eachtraining phase perhaps best represents the ef-fects of training alone, that data point couldalso represent a day on which disruption wasunusually high, as a result of which it is im-

portant to consider the childrens performanceacross all of the subsequent testing sessions.

Finally, it should be noted that testing wasconducted under extinction, and no effort wasever made to maintain trained responses oncethe training criterion had been reached. Oncea mand or a tact repertoire had been trained, itwas never again reinforced (except in the case

of Olivia, who received tact training on thecube twice). Some of the data obtained in thisstudy may thus be attributable to extinction;such as Noahs poor performance on mand tri-als when mands had already been trained tocriterion, and all of the childrens failure tomaintain accurate responding over extended

periods of time. It is also possible that the emer-gent repertoire is more susceptible to extinc-tion than the trained repertoire, and that thiscontributed to the observation that in some

cases there was evidence that the untrained rep-ertoire emerged, but performance was poorerthan in the trained repertoire or declined overtime (Tristan and Emmas performance onmand trials following tact training, and Olivias

performance on mand trials following the sec-ond round of tact training).

In conclusion, Skinners (1957) analysis ofverbal behavior suggests that mands and tactsare functionally distinguishable verbal oper-ants, each requiring a unique learning history

to be established. However, the analysis doesnot necessarily imply that one is typically ac-quired without the other, as a number of con-ditions could potentially give rise to mand-tacttransfer. Additional research on mand-tact in-


16/16


dependence in typically developing children isneeded to shed light on seemingly discrepantfindings in the small existing literature. Futureresearchers should attempt to isolate the vari-

ables that may give rise to mand-tact transfer.Experimental identification of such variableswould not only further our understanding oflanguage development from a behavior-ana-lytic perspective, but might also result in theidentification of potential language interven-tion techniques.

REFERENCES

Barnes-Holmes, D., Barnes-Holmes, Y., &Cullinan, V. (2000). Relational frame theoryand Skinners Verbal Behavior: A possiblesynthesis. The Behavior Analyst, 23, 6984.

Carroll, R. J., & Hesse, B. E. (1987). The ef-fects of alternating mand and tact trainingon the acquisition of tacts. The Analysis ofVerbal Behavior, 5, 5565.

Egan, C., & Barnes-Holmes, D. (May 2004).Assessing the independence of function

between mands and tacts. In C. Egan(Chair), Contextual effects on verbal com-munication. Symposium conducted at the

30th annual convention of the Associationfor Behavior Analysis, Boston, MA.

Goyos, C. (2000). Equivalence class formationvia common reinforcers among preschoolchildren.Psychological Record, 50,629654.

Hall, G., & Sundberg, M. L. (1987). Teachingmands by manipulating conditioned estab-lishing operations. The Analysis of Verbal

Behavior, 5, 4153.Hayes, S. C., Barnes-Holmes, D., & Roche, B.

(2001). Relational frame theory: A post-

Skinnerian account of human language andcognition.New York: Plenum.

Horne, P. J., & Lowe, C. F. (1996). On the ori-gins of naming and other symbolic behav-ior.Journal of the Experimental Analysis of

Behavior, 65, 185241.Horne, P. J., Lowe, C. F., & Randle, V. R. L.

(2004). Naming and categorization in youngchildren: II. Listener behavior training.

Journal of the Experimental Analysis of Be-havior, 81, 267288.

Lamarre, J., & Holland, J. G. (1985). The func-tional independence of mands and tacts.

Journal of the Experimental Analysis ofBehavior, 43, 519.

Lowe, C. F., Horne, P. J., Harris, F. D. A., &Randle, V. R. L. (2002). Naming and cat-egorization in young children: Vocal tacttraining.Journal of the Experimental Analy-

sis of Behavior, 78, 527549.Michael, J. (1988). Establishing operations and

the mand. The Analysis of Verbal Behavior,6, 39.

Michael, J. (1993). Establishing operations.The Behavior Analyst, 16, 191206.

Mish, F. C. et al. (Eds.). (1993). Merriam-Websters collegiate dictionary (10th ed.).Springfield, MA: Merriam-Webster.

Nuzzolo-Gomez, R., & Greer, R. D. (2004).Emergence of untaught mands or tacts ofnovel adjective-object pairs as a functionof instructional history. The Analysis of Ver-bal Behavior, 20, 6376.

Saunders, R. R., Drake, K. M., & Spradlin, J.E. (1999). Equivalence class establishment,expansion, and modification in preschoolchildren.Journal of the Experimental Analy-

sis of Behavior, 71, 195214.

Sidman, M., Willson-Morris, M., & Kirk, B.(1986). Matching-to-sample procedures andthe development of equivalence relations:The role of naming.Analysis and Interven-tion in Developmental Disabilities, 6, 119.

Sigafoos, J., Doss, S., & Reichle, J. (1989).Developing mand and tact repertoires in

persons with severe developmental disabili-ties using graphic symbols. Research in

Developmental Disabilities, 10, 183200.Sigafoos, J., Reichle, J., Doss, S., Hall, K., &

Pettitt (1990). Spontaneous transfer ofstimulus control from tact to mand contin-gencies. Research in Developmental Dis-abilities, 11, 165176.

Skinner, B.F. (1957). Verbal behavior. Acton,MA: Copley.

Twyman, J. (1996). The functional indepen-dence of impure mands and tacts of abstractstimulus properties. The Analysis of Verbal

Behavior, 13, 119.

Emergence of Mands and Tacts of Novel Objects among Preschool Children

Documents

Transcript of Emergence of Mands and Tacts of Novel Objects among Preschool Children