ogden r. university behavior...

The Behavior Analyst 1996, 19, 199-210 No. 2 (Fall)

The Four Free-Operant FreedomsOgden R. Lindsley

University of Kansas and Behavior Research CompanyThis article reviews early free-operant conditioning laboratory research and applications. The sel-dom-mentioned four free-operant freedoms are described for the first time in detail. Most currentbehavior analysts do not realize that the freedom to form responses and the freedom to speedresponses were crucial steps in designing free-operant operanda in the 1950s. These four freedomswere known by the laboratory researchers of the 1950s to the point that, along with operanda design,Sidman (1960) did not feel the need to detail them in his classic, Tactics of Scientific Research.The dimensions of freedom in the operant were so well understood and accepted in the 1950s thatmost thought it redundant to use the term free operant. These issues are reviewed in some detailfor younger behavior analysts who did not have the opportunity of learning them firsthand.Key words: fluency, free operant, frequency, operant conditioning, rate of response

This paper is dedicated to Israel Gol-diamond, my close friend and belovedcolleague, who succumbed to multiplecancer on November 19, 1995. Al-though we seldom met more than oncea year, we were bonded by a mutualrespect for measurement and the free-operant method. I called him "Is" ' andhe called me "Og," and our relation-ship went back 43 years. Is understoodthe power of the free operant over dis-crete trials and used it, creatively, inboth laboratory and clinic from the1950s on. Along with other pioneers inbehavior analysis, Is regretted the factthat the less sensitive controlled oper-ant later dominated research in bothlaboratory and applied behavior anal-ysis. Eight of the 11 doctoral disserta-tions advised by Goldiamond in the1980s and 1990s were highly creativefree-operant designs. Topics rangedfrom symbolic aggression in pigeons(Andronis, 1983), through head bang-ing in pigeons (Layng, 1994), to sched-ule-induced defecation (Rayfield, Se-gal, & Goldiamond, 1982).

In the early days we both lived inthe academic fast lane. Rushing tomeetings, staying too late, never get-

Address correspondence to the author, 366North 1600 Road, Lawrence, Kansas 66049-9199 (E-mail: [email protected]).

Most of his students called him "Izzy," butto me he was always "Is." Certainly not one towear a diminutive! And, he always respondedrapidly in good humor when I called him "Is."

ting enough sleep, always perfection-istic, always trying to do too much.Sometimes I thought we were trying tomake up for the time we spent in theservice during World War II. This wenton until Is had his accident. Drivingback to Chicago from staying too latetalking to graduate students after an in-vited colloquium address at the Uni-versity of Illinois, Is demolished hisChecker car. This made him a paraple-gic and me a diesel Mercedes driver.Both my father and my brother hadbeen killed in speeding automobiles,but Is survived his crash and went onto do still more wonderful things.We will all miss the spark of his

laugh and stories. I can close my eyesand see and hear Is say, "Have I toldyou about the two boys who wereasked by their father, 'Who pushedouthouse off high cliff into PacificOcean?"' I can also hear the equal joyand excitement as he related the suc-cessful history of one of his creativenonlinear clinical behavior analysiscases. Life and science to Is were onebig grand adventure. If you couldn'thave fun doing it, it wasn't worth do-ing.

Is is now gone, and the lucky onesamong us have fond memories of himin addition to the record he left: hisextensive, high-quality research publi-cations. Even more important, he grad-uated over a dozen skillful, well-

199

200 OGDEN R. LINDSLEY

trained behavior analysts who are car-rying on the Goldiamond zest, researchtraditions, and life-style adventures.

THE FREE OPERANT

In separate telephone conversationson July 5, 1995, neither Charles Cata-nia, Jack Michael, nor Scott Wood, allexperts on the writings of B: F. Skinner,could recall that Skinner wrote thetermfree operant. All remembered thatFerster (1953) used the term in an earlyPsychological Bulletin article. I sug-gested to Catania, Michael, and Woodthat Skinner considered all operants tobe free; hence, free operant would beredundant. All experts agreed that mysuggestion was probably the case.

"Free-responding" is in the index ofTactics of Scientific Research. Sid-man's (1960) description follows.

The experimental situations that I have used forillustration all share at least one important fea-ture: the experimental organism is free to re-spond at any time. There are no harnesses torestrain the animal forcibly; the lever is neverwithdrawn from the experimental space to pre-vent the subject from responding at times thatwould be inconvenient for the investigator's the-ory. The only restrictions placed upon the sub-ject's recorded behavior are those inherent in thelaws of behavior. This is called a free-respond-ing situation. (p. 409)

In the 1950s and early 1960s, sev-eral laboratory researchers began touse the term free operant to differen-tiate situations in which the learner wasfree to make more than one responseto each stimulus from the controlledoperant, in which trials were used andthe learner responded only once toeach discriminative signal (Ferster,1953). Because many operant condi-tioners were beginning to do single-tri-al research, the others sometimes used"free operant" in the titles of their ar-ticles (Barrett, 1962; Creed & Ferster,1972; Ferster, 1953; Levin & Stigall,1965; Lindsley, 1960, 1963b; Lovitt,1967) in attempts to keep rate of re-sponse alive. These researchers usuallydescribed the power of the free operantin their method sections. For example,

We see in the method of free-operant condition-ing probably one of the most rigorous tech-niques yet developed by experimental psychol-ogy for the development, maintenance, modifi-cation and analysis of acquired motor behaviorin an experimental setting. B. F. Skinner per-fected this method, and he and his collaboratorshave been using it successfully for the pasttwenty-five years in analyzing the behavior oflaboratory animals. In application an animal isplaced unfettered and alone in a small enclosurewhere he is free to make any response at anytime-hence the term "free." If the animal op-erates a small lever, wheel, key, plunger or sim-ilar device, he is promptly rewarded or rein-forced-hence the term "operant." Throughvarying the nature and conditions of the rein-forcement, complex behaviors have been devel-oped and measured which are similar to sym-bolic behavior, "superstition," time-telling,counting, fear, anxiety, competition, coopera-tion, and so on.... The free-operant method canbe used, with very little modification, to measurethe behavior of any animal from a turtle to nor-mal genius. (Lindsley, 1956, p. 118)

It was also the custom during the1950s and 1960s for operant condition-ers to present the advantages of thefree operant over the use of trials (i.e.,controlled or restricted operants) intheir method sections. Here again,

Over the past 30 years Skinner has stressed thevalue of the free operant, wherein the subject isfree to respond at any time. The rate of occur-rence of this response is the primary datum. Useof the free operant dispenses with burdensometime-consuming trials and eliminates the con-founding stimulation of the trials themselves.Also subtle variations in the behavior beingstudied, which could occur between trials, arefully recorded in the continuous monitoring ofthe free-operant technique. (Lindsley, Hobika, &Etsten, 1961, p. 937)

Before adopting the free operant andrate of response as its measure, manyresearchers had used experimental psy-chological methods employing trialsand either probability of occurrence(percentage), amplitude, latency, or tri-als to extinction, the four Hullian re-sponse measures (Hull, 1943). This ex-perimental history produced an acuteawareness of the vastly superior sen-sitivity of the free-operant method andrate of response as a measure. For thisreason, concern was raised when someoperant conditioners began using sin-gle-response trials, with percentagesand latencies as the response measures.

FOUR FREEDOMS 201

The tendency to revert to trials rath-er than the free operant was verystrong in regular and special educationand in child development applications.In the University of Washington ChildBehavior Laboratory School in the late1950s, Bijou and his students usedplunger operanda and cumulative re-corders with the children in their free-operant laboratory, but used trials andpercentage correct with the same chil-dren in their nearby classrooms. Bijouwas Spence's2 second doctoral student,and from 1941 through 1955 was aconfirmed Hull-Spence behavioristand experimenter (Krasner, 1977). Be-ing a Hullian, he turned to Sears' mod-el for research with children and start-ed with a ball-drop response (Bijou,1955). Later Bijou replaced the dis-crete-trials ball-drop with a free oper-ant in which the child pushed the noseon a clown's face, but he still used anevent recorder for stimulus and re-sponse occurrences rather than thefree-operant cumulative response re-corder (Bijou, 1957). My laboratoryvisitors' book shows that on April 1,1957, Bijou spent the day at my Har-vard Medical School Behavior Re-search Laboratory in MetropolitanState Hospital, Waltham, Massachu-setts. There he saw the plungers that Ihad designed for human free-operantwork that he later ordered from RalphGerbrands Company for use in his ownlaboratory at the University of Wash-ington. The plunger operanda wereused in some of Bijou's later labora-

Kenneth Spence, professor of experimentalpsychology at Iowa, popularized Hull's mathe-matico-deductive theory approach. Spence had amap of North America on his office wall withpins stuck in the locations of the universitieswhere his doctorates were employed. The Sp-ence-Hullians controlled experimental psychol-ogy and its journals. I was taught by Greg Kim-ball, a Spence PhD, when I was a graduate stu-dent at Brown. At that time I was a graduatestudent in physiological psychology and a Hul-lian when it came to learning. I belonged to the"postulate of the month club" and got dittoedcopies of upgrades to the theory from Hull's of-fice at Yale. I realized the theory was doomedwhen I couldn't keep up with the upgrades.

tory research (Bijou, 1958). From theIowa Child Welfare Station, Bijou alsopicked up the method of "percentageof time observed on task," the tradi-tional and entrenched child develop-ment measure (Goodenough, 1928).The tendency to drop the free oper-

ant in human applications was sostrong that even Skinner used con-trolled operants in his applications withteaching machines (Skinner, 1958). Idid not support Skinner's work withteaching machines for three reasons.First, there was no opportunity to re-peat a response in the presence of itsstimulus. Second, the response mea-sure was percentage of frames correct.Third, there was no provision for ex-ternal reinforcement. Teaching ma-chine proponents said that getting theanswer right was the reinforcement. Atthat time, I would say, "If getting itright is ample reward, everybody in theworld would be getting more and moreright, just at different rates." True,there was a "free" aspect to teachingmachines and programmed instruction.Learners were free to answer the ques-tions at their own pace and free to pres-ent the questions, but they were notfree to respond more than once to eachquestion presentation.

Others started to record latencies orinterresponse times in place of rate ofresponse. Some took the position thatlatencies or interresponse times mea-sured the same thing that rate did, butdid so more precisely; they believedthat rate and cumulative records werenothing more than a primitive way ofrecording latencies. This is really aneo-Hullian position, because it im-plies that rate of response, latency, in-terresponse time, duration, and ampli-tude 'are just alternative ways of mea-suring response strength. And responsestrength is what you really are after.Response strength is an interveningvariable, an hypothesis, and it is neverdirectly measured. Later, matching-to-sample and stimulus equivalence re-search became popular, and these re-searchers also rejected the free operantand used trials. Because trial research


gained so much favor among behavioranalysts (the new name for operantconditioners), cumulative records grad-ually all but disappeared from theJournal of the Experimental Analysisof Behavior' (Barrett, 1987). They hadonly rarely been published in the Jour-nal of Applied Behavior Analysis.

Skinner regretted this noticeable de-cline in the use of cumulative record-ing (Skinner, 1976). He should alsohave regretted the decline of the free-operant method that fell from favor atthe same time. The decline of the free-operant method is probably why thecumulative record went out of favor. Acumulative record of a learner's re-sponses in a situation in which stimuliare presented in fixed order, at a timeselected by the experimenter, run by atimer, in which the learner can makeonly one response to each stimulus,and cannot skip or correct or repeat re-sponses, would record the experimen-ter's behavior, not the learner's re-sponse. In these controlled conditionsa cumulative record added little value,and a table or chart of percentage ofcorrect responding over trials was ad-equate for such research.

I closed my human free-operant be-havior research laboratory at HarvardMedical School and Metropolitan StateHospital and accepted a faculty ap-pointment in the University of KansasSpecial Education Department to ad-vocate using rate of response in edu-cational research and classroom teach-ing (Lindsley, 1966, 1971, 1972). I didthis because most behavior analystsworking in educational settings usedpercentage correct as the measure ofclassroom academic performance. Thepreponderance of percentage measuresin application and the decline of ratemeasures in the basic laboratories and

' One of the main reasons for establishing theJournal of the Experimental Analysis of Behav-ior was to provide a place to publish cumulativerecords. In the 1950s both the Journal of Ex-perimental Psychology and the Journal of Com-parative and Physiological Psychology routinelyrejected articles that contained cumulative re-cords.

journals required drastic action4 if rateof response was to survive. Obviouslylaboratory research and commentariesabout the glories and power of the freeoperant and cumulative recordingfailed to stem the tide of regression tothe four Hullian response measures. Atthe time one occasionally heard a free-operant supporter ask a friend who hadgone back to trials research "Did yougo back to Hull, or to Hull and back?"

This lack of interest and use of rateof response and cumulative records re-sulted in little writing about them andvery little study of the advantages andfeatures of the free operant. The freeoperant or free responding do not ap-pear in the index to Strategies and Tac-tics of Human Behavioral Research(Johnston & Pennypacker, 1980). In arecent count of pages devoted to thetopic of the free operant in 17 booksauthored by Skinner and four methodpublications by other authors,5 I talliedonly 12 of over 5,000 pages describingthe free operant.6

FOUR FREE-OPERANTFREEDOMS

For this article, I have recalled andspelled out the details of what the zeal-ous laboratory proponents of free-op-erant research knew in the 1950s and1960s. In fact, operant freedoms ex-ceed the commonly known freedom topresent stimuli and the less well-known freedom to repeat responses.Other freedoms include little-knownbut equally important freedoms to formand to speed responses. These fourfreedoms are crucial to developing andmaintaining fluent performances.

4 This is not to imply that going from Cam-bridge to Kansas City is drastic; rather, closinga really productive laboratory is drastic.

The four method books by other authorswere Keller and Schoenfeld (1950), Sidman(1960), Ferster and Culbertson (1982), and John-ston and Pennypacker (1980).

6 A table of the counts of pages devoted to thefree operant, shaping, chaining, and functionalrecording is available from the author.

FOUR FREEDOMS 203

Free to Present Stimuli

The learner is always ideally readyand is never caught napping or off bal-ance when free to present stimuli; thefreedom to "self-pace" or "self-pres-ent" is the best known operant free-dom. The destructive effect of externalpacers on the development of fluentperformances, and the constructive ef-fects of freedom to stop, start over, cor-rect, and skip, however, are less wellknown so are touched on here in a littlemore detail. My classroom use of SAF-MEDS ("say all fast a minute each dayshuffled") fluency cards showed thatthe learner's frequency drops to abouthalf the self-held frequency when an-other student held the learner's cards,exposing the next card right after thelearner responds. Self-presenting wastwice as fast as partner presenting.The freedom to present the stimuli

fosters response rhythms (e.g., pointingto the next frame on a practice sheet,sliding the top card off the SAFMEDSdeck and exposing the next card ques-tion). Response rhythms are crucial fordeveloping high-fluency frequencies. Itried three forms of external pacing tospeed up responding: an external met-ronome clicking at 60 per minute, atelephone busy signal at 72 buzzes perminute (learners dialed their own num-ber), and a partner tapping a pencil ona table, regularly, at 60 taps per min-ute. All of these regular, clock-like, ex-ternal pacing attempts failed. Mostlearners could not keep up with thepacing and "clutched" to a stop after10 or 15 cards. Regular external pacingbroke the learner's natural rhythm andrattled the learner trying to build flu-ency.

Alfredo Lagmay, a 1950s graduatestudent of Skinner's from the Philip-pines, made a classic study of pacing,in which reinforcement was contingentupon pecking at a steady rate. Whenthe schedule would otherwise havegenerated a higher rate, the pigeonwould break through and "get rid ofextra responses" in a sudden burst(Skinner, 1983, p. 38). Here, from the

pigeon laboratory, is more evidencethat learners need to repeat at their ownrhythmical pattern, and cannot long en-dure regular external pacing.

Irregular, rhythmical, external pac-ing by a partner tapping the table witha pencil at a speed slightly faster thanthe learner's responding worked. Thetapper matched the tapping to therhythm of the learner's responding,varying the speed from time to time,but always trying to lead the learner upto 60 responses per minute as rapidlyas possible without losing the learner.This rhythmical, synchronized pacingis similar to what a well-trained cox-swain does in pacing a rowing crew. Italso is what skilled direct-instructionteachers do when they pace their learn-ers to "firm " the performance.Freedom to stop and start over is an-

other presentation freedom, which isalmost always used by precision teach-ers in fluency building. Quite often astudent will "clutch" in the middle ofa 1-min timing, especially during finaltimings with the teacher or professorfor a course grade. That is why preci-sion teachers usually permit three orfour timings in a course grade check-out and use the timing with the highestfrequency of correct responses (withsome students this is the first timingand others it is the last of the four). Itis also standard practice to permit asecond check-out a day or a week laterif the learner wishes to practice somemore for a higher grade.Freedom to skip can be permitted if

the teacher chooses. In building fluen-cy with SAFMEDS, learners reach theinstructional frequency aims soonerwhen they start out at high overall fre-quencies, showing steeper accelera-tions in frequency. They start at highspeed (60 per minute) and low accu-racy (one hit for every 20 misses). Thisprocedure requires high curricularcourage from both teacher and student,but it produces the steepest learning.To foster these high beginning frequen-cies, I encourage learners to say "go"to cards for which they cannot give animmediate answer. When learners hes-


itate and then say a correct answer,they learn to grope and to pause, whichdefeats fluency development. Whenlearners mumble and say "I don'tknow," that also takes time and tendsto emotionally depress them and theygo even slower. I urge SAFMEDSlearners to say "go" for every cardthat the answer does not jump rightout. The shortness of "go" does notslow responding. The emotional effectof "go" speeds or hustles responding.The single syllable of "go" does notbreak the learner's "slide-see-say"rhythm. In this sense "go" responsesare carefully designed, highly speci-fied, and approved skips.

Free to Form Responses

Very little has been written on thedesign of manipulanda. It was essentialto select a class of responses that werenot identical in form and that yieldeda smooth and lasting variable-interval(VI) response rate. The learner wasfree to select the most comfortable re-sponse form and to vary that form dur-ing experimental sessions to overcomeboredom and fatigue. The wide re-sponse class also permitted differentindividuals to use their own, unique,most comfortable response form. Readwhat Reese wrote about operanda inthe "Free Operant Method" section ofher popular textbook, The Analysis ofHuman Operant Behavior (1966).

The gerbil in Figure 2 is pressing one kind ofmouse lever. When the lever is depressed a cer-tain distance, it operates a microswitch whichdefines the response. The gerbil may press withether paw or both paws, or he may climb on thelever or straddle it. Only those behaviors-butall of those behaviors-which operate the mi-croswitch are defined as lever-pressing re-sponses. (p. 5)

Many weeks (sometimes years) werespent in the early 1950s designing op-eranda that would produce smooth,even rates of response on the 1 -min VIreinforcement schedule that producedregular stable responding. Only the fewof us who actually designed operandafor a new species were acutely awareof the uneven rates of response pro-

duced by operanda that narrowed theform of the response too far. With toonarrow a response form requirementfatigue set in early, and sometimeschoppy rates would occur on a sched-ule that was designed to produce evenrates.

For example, in designing an oper-andum for beagle dogs, I (Jetter, Lind-sley, & Wohlwill, 1953; Lindsley,1957) at first tried a lever very muchlike a rat lever only larger and at anappropriate height. It promptly pro-duced a chopped rather than a smooth,even VI rate, because the dogs chewedthe lever at times rather than pressingit. To rule out chewing,7 I built a panelat the end of a shallow tube. The dogsput one paw into the tube and pressedthe panel at the end. This eliminatedchewing entirely, but produced an un-even VI rate as the dogs' toenails gotsore and sometimes caught on the edgeof the tube. Next, I mounted a squareoperandum panel hinged at its top atthe right side of the chamber.8 Thiseliminated both chewing and nail ten-derness, but the rate of response wasstill uneven, because the dogs shiftedtheir weight and their right paw wouldtire. Next, I mounted the operandum inthe middle of the end of the chamberand the dogs alternated either frontpaw, both paws, or their muzzles. Thisoperandum produced the widest re-sponse class so far, and the rate of re-sponse was now almost smooth withthe VI schedule. Occasionally a dogwould paw at the top of the panel,however, and this pawing would notoperate the microswitch because thepanel was hinged at that point andwould not depress. This added occa-sional chop in the record that shouldbe smooth, so I hinged the operandum

7In those days one never ruled out chewingby ruling out dogs that chewed. All laboratoryfree-operant conditioners were proud of neverhaving to discard an animal. That proved ourbehavioral power, knowledge, and control. Onewould have been ashamed to be caught discard-ing subjects!

The panel had a Plexiglas window behindwhich stimulus lights were mounted.

FOUR FREEDOMS 205

panel on internal struts about 24 in.above the panel on the outside of thechamber end to produce the final suc-cessful design. The top of the paneland the bottom moved the same dis-tance to trip the microswitch responsecounter, and right front paw, left frontpaw, and muzzle presses were recordedat the top as well as at the bottom andsides of the panel. Now the 1-min VIschedule produced a smooth, even re-sponse rate over a period of severalhours, and we finally had an appropri-ate dog operandum that gave the bea-gles freedom to form their responses.The Ferster-Skinner pigeon key fi-

nally marketed by Ralph Gerbrandsalso went through many versions be-fore it recorded the proper responseclass width. This key permitted thebirds to peck in many different formssuch as front tap, side swipe, andslightly open peck; each bird used itsown particular response form. The keyeliminated chatter (switch closures un-related to responding) and recorded awide range of peck forces. The keyalso had to be durable enough to op-erate at six million responses permonth over periods as long as 2 yearswithout failure (Ferster & Skinner,1957, p. 17). Most important, the finalversion of the key produced smooth,even responding at high rates for ses-sions as long as 15 hr. The Anger9 ratlever was also designed to produceeven responding on a VI schedule, andrats could respond with their left paw,right paw, and even their chin. The in-destructible Lindsley plunger operan-dum'° for humans permitted partici-pants to use either or both hands in re-sponding at even rates for 7 hr on VIschedules.

It also helps to let the learners maketheir own abbreviations to lift the ceil-ing on rate imposed by the length ofthe answer words during fluency build-ing with practice sheets using the

9 Douglas Anger was a graduate student ofSkinner's and a Junior Fellow at Harvard." The Lindsley plunger was also manufac-tured and sold by Ralph Gerbrands Company.

point-see-say channel (Lindsley, 1994).For example, if the correct answer to acard was "increase," then "inc," "in-cre," or "i" could be accepted as anabbreviated answer, lifting the answerceiling and permitting a higher answerfrequency. If the channel was point-see-write, the learner could abbreviatethe answer with an up-arrow in placeof writing or abbreviating "increase."This freedom to form the responseclass lets the learners invent their ownabbreviations that are often superior toany the teacher might devise.Freedom to make your own is the

ultimate freedom to form." For the first3 years of using SAFMEDS in my uni-versity graduate classes, I had studentsmake their own deck of cards fromblank index cards as the first week'sassignment. The cards were made froma list of questions (front of card) andanswers (back of card) that were dis-played on an overhead screen in thefirst class. Student objection to theviewing problem caused me to pass outa sheet with columns of card fronts andbacks. I believed strongly that the stu-dents valued their SAFMEDS decksmore when they made the deck them-selves. No student ever lost a self-made SAFMEDS deck, but every se-mester several students lost their deckwhen I succumbed to student pressureand provided ready-made SAFMEDSdecks.

This freedom to make one's own in-cludes the entire construction of learn-ing materials and the choice of content.Given instructions, most students canselect a textbook section, choose theirown important topic points, write ques-tions and answers for these points, con-struct their own SAFMEDS deck, anduse SAFMEDS on their own. With thefreedom to make one's own, studentslearn how to develop fluent perfor-

" Harold Kunzelmann told me to be sure tomention "free to make your own" when I re-cently telephoned him to ask permission to citeHarold's extinction by public education in an ar-ticle dedication. Harold is now gainfully self-employed in the private sector.


mances, and at the same time acquirethe skills to become fluent in most fu-ture learning tasks. That is the ultimatein personal freedom to form.

Free to Repeat Responses

The freedom to repeat responsesmany times to each signal was an im-portant feature of the laboratory re-search using free operants. Repeatingresponses permitted researchers to di-rectly and continuously record the lear-ner's degree of assurance to each signalin a discrimination experiment. Therewas no need for recourse to statistics,as required with trials that permittedonly one response per signal.We studied discriminations in the

early 1950s by using free operants withthe freedom to repeat rather than thecumbersome and insensitive trials.Skinner described beautifully the free-operant discriminations from studentprojects for his first graduate seminarat Harvard in 1950.

The students "shaped behavior through succes-sive approximation," reinforced on intermittentschedules, and brought the behavior under thecontrol of discriminative stimuli. William Mc-Gill, produced a rough "generalization gradi-ent" by intermittently reinforcing pecks on ayellow triangle and noting how fast the pigeonthen pecked triangles of other colors during ex-tinction. George Heise recorded a gradient forthe size of a spot on the key. When pecking aparticular size of the spot was reinforced, howfast would the pigeon peck other sizes? My lab-oratory assistant, Sam McLaughlin, used thesame method to test a pigeon's ability to tell thedifference between the figures used on licenseplates. He built up a high rate of pecking to thefigure 6 and then, during extinction, replaced the6 for short periods with other figures from 0 to9. The pigeons pecked at rates which reflectedwhat would be called in the human case, "per-ceived differences between the figures." (Skin-ner, 1983, pp. 6-7)

Many current researchers have for-gotten or never knew how to use thefree-operant measurement in the waythat Skinner described. They believethat researchers must use trials withonly one response per stimulation tostudy discrimination, matching to sam-ple, or stimulus equivalence. Theyhave forgotten the fantastically high

sensitivity of Blough's dark adaptation(1956) and generalization experiments(1957, 1967) with pigeons.

Skinner also eliminated the freedomto repeat answers, to self-correct, andto skip questions when he designedteaching machines. Speech patholo-gists know that in improving pronun-ciation there is a stage in which learn-ers cannot pronounce correctly, but canrecognize their own incorrect pronun-ciation as they hear themselves speak-ing. Speech pathologists find it valu-able to permit learners to immediatelyrepeat and self-correct their incorrectpronunciation. Programmed instruc-tion, following Skinner's example, per-mitted only one response per question.Precision teaching practice sheets'2 andSAFMEDS also permit only one re-sponse to each stimulus, althoughsome precision teachers encourageself-correcting in the point-see-saylearning channel.

Free to SpeedThe freedom to speed in repeating

responses to a single stimulus was veryimportant to early laboratory free-op-erant research. Researchers spentmonths and years designing operandathat operated faster than the learnerscould move to prevent putting an op-erandum ceiling on the response rate.Most behavior analysts do not realizethat all the charts published in Skin-ner's classic, The Behavior of Organ-isms (1938), had a ceiling at 60 re-sponses per minute.In the first experiments with this method a nee-dle attached to the lever-arm dipped into a smallcup of mercury. When the lever was movedslowly there was a tendency for the contact tochatter, and this was corrected by inserting intothe circuit to the recorder a device which madeit impossible for a second contact to be recordedwithin, say, one second. (p. 60)

This false recording apparatus ceilingwould not permit recording responsespeeds above 60 per minute (one per

12 Practice sheets have also been called skillbuilders (Beck, Conrad, & Anderson, 1995) andfluency sheets (Binder, 1993).

FOUR FREEDOMS 207

second). This limitation bothered Skin-ner, so in later experiments he used acommercial mercury tube switch thatdid not require the 1-s limiting delay.The Ferster-Skinner pigeon key and

Gerbrands recorder could eventuallyfollow frequencies as high as 15 persecond (900 a minute), well above thehighest pigeon frequencies of 800 perminute. In contrast, during initial at-tempts to apply fluency research meth-ods to computer-assisted instruction,most of the early personal computersrequired as much as 3 s to refresh thescreen and display a graphic image forthe next question (Silverman, Lindsley,& Porter, 1991). This limited the lear-ner's response frequency to less than20 per minute, which was not fastenough to develop true fluency.

FREE-OPERANTRESEARCH DESIGN ANDTRAINING PROCEDURES

Two different procedures developedin the 1950s are still used today by afew of us to design both laboratory andapplied free-operant research. Both arealso useful in training students to thinkfree operant. The first I called "exter-nalizing the behavior." The second Icalled "freeing up the operant." Littlewas written about these powerful pro-cedures because they were common-place in the laboratory jargon of the1950s and thus do not appear in Sid-man (1960), Johnson and Pennypacker(1980), or any of the laboratory man-uals and college texts that I have read.I used both in my workshops and grad-uate classes throughout my teachingcareer, and they successfully overcamemany students' ingrained tendencies toattempt complete control in experi-mental design. These powerful proce-dures are offered here in the hope thatthey may facilitate free-operant re-search and practice.

Externalizing the BehaviorI can still hear Skinner telling me,

"You externalized empathy!" when helooked at the cumulative records of

one of my psychotic patients' respond-ing to feed a hungry kitten (Lindsley,1963a). "Externalizing the behavior"was a common phrase in the 1950s and1960s among those of us who werebringing socially important behaviorinto the free-operant laboratory for thefirst time. At the University of Kansasduring the late 1960s, we successfullyadvised parents, teachers, and studentsin Special Education on how to con-duct self-modification projects. Theseprojects included not only stopping hit-ting, swearing, and smoking and re-ducing weight, but also stopping hit-ting, swearing, smoking, and eatingurges. The urges were externalized byhaving the clients push a button on onewrist counter each time they did it, anda button on second wrist counter eachtime they had an urge to do it (Duncan,1969, 1971).One of the most telling indicators in

the 20-timing assessment battery of theHaughton Learning Center (Napa, Cal-ifornia) is "see-say" nouns. Prospec-tive students are handed a Veeder-Roothand tally counter that they click eachtime they say the name of an object inthe room as fast as they can during a1-min timing. Adjectives are notcounted (e.g., red binder, green binder,blue binder, yellow binder). Most firstgraders start at about 20 nouns perminute, and after several weeks ofpractice are up to their aim of 50 to 70nouns said and counted per minute.The next step is to time think-saynouns recalled and visualized fromone's bedroom at home, from thekitchen, or from the garage. The laststep is to time think-say nouns fromcategories such as vegetables, animals,or minerals. Because students enjoysee-say nouns more than any otherpractice task at the center, it is usuallyused as a warm-up exercise each day.Recently, an 18-year-old student with auniversity water polo scholarship wasonly able to see-say nouns at 31 perminute in his assessment. This college-bound man desperately needed practicein basic description and word use (E.


F Haughton, personal communication,July 14, 1996).Good chess teachers ask their learn-

ers to think out loud; with theirthoughts externalized, the teachers canmore easily advise their learners.Morningside Academy (Seattle) usesWhimbey's (1989, 1995) think-aloudproblem solving (TAPS) to teach prob-lem solving. These methods greatly fa-cilitate teaching the more complexreading analysis and problem solving(Kent Johnson, personal communica-tion, June 4, 1996), and they are spe-cial cases of externalizing the operant.Externalizing the behavior is still auseful exercise in teaching students touse the free operant in place of the dis-crete trials of the controlled operant.Learners and teachers prefer the freeoperant over a controlled operant (oreven worse, rating scales) once theyknow how to externalize it.

Freeing Up the Operant

Perhaps even more useful in designand training is freeing up the operant.This became especially important inmy years teaching graduate courses inspecial education. There, due to the in-fluence of physical rehabilitation, thecontrolled-trial procedures were in-grained in almost all remedial pro-grams. Experienced teachers who cameback to graduate school for certifica-tion in special education had great dif-ficulty thinking of how they couldmake teaching a pupil to catch a ballinto a free operant. Three examples offreeing up ingrained controlled oper-ants follow.

Catching a ball is usually taught byslowly tossing a large ball at a shortdistance to the learner. One might evenstart with rolling the ball across thefloor. The ball is retrieved by the teach-er and another toss is made; the ball isseldom tossed more than once per min-ute. The size of the ball and the dis-tance are selected to make almost allof the catches successful. Gradually theball size is decreased, the distance islengthened, and the speed is increased.

Usually only one ball of each size isavailable, so the catching is not timedand there are usually no more than 30catches in the half hour of class time.Full attention of one aide is required totoss and retrieve the ball. One methodof freeing up the operant is to get abushel basket of spent tennis balls fromthe local tennis club. Set a 1-min timer,and at a distance of about 15 feet throwballs at the learner at a frequency of 60balls per minute.'3 Most will not becaught, but one or two will stick be-tween the learner's arms and body-those are caught! Count catches andmisses and have 10 or 15 1-min ball-catching trials per day. In the first30-min 15-trial day, the learner maycatch 35 balls and miss 865 balls froma total of 900 (15 times 60 tosses).These 900 tosses in one free-operantday give 30 times more practice thanthe 30 tosses in a 1-min discrete-trialsday. Learners count and chart theircatches and misses each day.

Walking a balance board is usuallytaught by having the learner walkdown a raised plank (6 ft long, 6 in.wide) on the floor. An aide may at firsthold the learner's hand to guide him orher. Sometimes the learner is walkedback to the front of the plank for an-other trial. Sometimes the learner isturned around and walked back downthe plank for a second trial. Slips offthe plank are counted as errors. Thisoperant is freed up by making a plankin a circle (6 ft inside diameter) (orpainting one on the floor). The aidemay stand in the center and time 1 -minwalks and count correct steps on theplank and misses off the plank aloud,with the learner also counting out loud.Ten or 15 1-min timings per day canbe practiced and the hits and missescharted by the learner.

Saying "good morning" meaning-fully is usually taught with one trial aday as the learner enters the classroom.

1' It is even better when another student tossesthe balls, because then two students are learningand the teacher could coach three or four ball-tossing, ball-counting teams at once.

FOUR FREEDOMS 209

Some more enlightened teachers mayhave the learner greet every other childin the room, and have each greeteecount whether he or she was greetedwarmly and meaningfully or not. Thatwould give as many learning opportu-nities each day as children in the room.The "good morning" operant is stillnot really free. The "good morning"operant could be totally freed by form-ing the children in a large circle andhaving the greeter child go around thecircle greeting each child in the circle,and not being able to go on to the nextchild until the greetee signals with hisother arm "warm and meaningful."The number of warm and meaningfulgreetings per minute is the hit frequen-cy for the learning greeter.

IMPORTANCE OF THE FOURFREE-OPERANT FREEDOMS

FOR THE FUTURESome of us never forgot the power

and greater sensitivity of the free op-erant and taught our students free-op-erant research design and clinical andeducational practice. Goldiamond di-rected eight free-operant dissertationsduring the 1980s and early 1990s (e.g.,Andronis, 1983; Layng, 1994) andonly three discrete-trials dissertationstesting the yes-no design of traditionalsignal-detection research. Other youngresearchers are aware of how they cando stimulus equivalence research usinga free operant (Rosales-Ruiz & Fitzim-mons, 1996).One purpose of this article is to cau-

tion those researchers who believe thatthey can study fluency systems and oneor more of their products (Lindsley,1995) using controlled-operant re-search designs, with discrete trials andlatencies as the response measure. An-other major purpose is to open the richfield of free-operant laboratory re-search and the even richer field of free-operant clinical and educational prac-tice to the next generation of behavioranalysts. You are empowered and arenow free to use your four free-operantfreedoms.

REFERENCESAndronis, P. T. (1983). Symbolic aggression bypigeons: Contingency coadduction. Unpub-lished doctoral dissertation, University of Chi-cago.

Barrett, B. H. (1962). Reduction in rate of mul-tiple tics by free operant conditioning meth-ods. Journal of Nervous and Mental Disease,135, 187-195.

Barrett, B. H. (1987). Drifting? Course? Desti-nation?: A review of research methods in ap-plied behavior analysis: Issues and advances.The Behavior Analyst, 10, 253-276.

Beck, R., Conrad, D., & Anderson, P. (1995).Basic skill builders handbook. Longmont,CO: Sopris West.

Bijou, S. W (1955). A systematic approach toan experimental analysis of young children.Child Development, 26, 161-168.

Bijou, S. W (1957). Methodology for an ex-perimental analysis of child behavior. Psycho-logical Reports, 3, 243-250.

Bijou, S. W. (1958). A child study laboratoryon wheels. Child Development, 29, 425-427.

Binder, C. (1993). Behavioral fluency: A newparadigm. Educational Technology, 33(10),8-14.

Blough, D. S. (1956). Dark adaptation in thepigeon. Journal of Comparative and Physio-logical Psychology, 49, 425-430.

Blough, D. S. (1957). Spectral sensitivity in thepigeon. Journal of the Optical Society ofAmerica, 47, 827-833.

Blough, D. S. (1967). Stimulus generalizationas signal detection in pigeons. Science, 158,940-941.

Creed, T. L., & Ferster, C. B. (1972). Space asa reinforcer in a free-operant environment.Psychological Record, 22, 161-167.

Duncan, A. D. (1969). Self application of be-havior modification techniques by teen-agers.Adolescence, 6(4), 541-556.

Duncan, A. D. (1971). The view from the innereye: Personal management of inner and outerbehaviors. Teaching Exceptional Children,3(3), 152-156.

Ferster, C. B. (1953). The use of the free op-erant in the analysis of behavior. Psychologi-cal Bulletin, 50(4), 263-274.

Ferster, C. B., & Culbertson, S. A. (1982). Be-havior principles (3rd ed.). Englewood Cliffs,NJ: Prentice Hall.

Ferster, C. B., & Skinner, B. F (1957). Sched-ules of reinforcement. New York: Appleton-Century-Crofts.

Goodenough, F (1928). Measuring behaviortraits by means of repeated short samples. TheJournal of Juvenile Research, 12, 230-235.

Hull, C. L. (1943). Principles of behavior. NewYork: Appleton-Century.

Jetter, W W., Lindsley, 0. R., & Wohlwill, F. J.(1953). The effects of irradiation on physicalexercise and behavior in the dog: Related he-matological and pathological control studies.Boston University Medical School, Final Re-


port, AEC Contract AT(30-1)1201. Micro-card number: NYO-4548.

Johnston, J. M., & Pennypacker, H. S. (1980).Strategies and tactics ofhuman behavioral re-search. Hillsdale, NJ: Erlbaum.

Keller, F S., & Schoenfeld, W. N. (1950). Prin-ciples of psychology. New York: Appleton-Century-Crofts.

Krasner, L. (1977). An interview with SidneyW. Bijou. In B C. Etzel, J. M. LeBlanc, & D.M. Baer (Eds.), New developments in behav-ioral research: Theory, method, and applica-tion (pp. 587-599). Hillsdale, NJ: Erlbaum.

Layng, T V. J. (1994). Programmed head-banging by pigeons: Cessation and return.Unpublished doctoral dissertation, Universityof Chicago.

Levin, R. H., & Stigall, T. T. (1965). Effect ofnonverbal social reinforcement on rate ofspeech in a free operant paradigm [Abstract].American Psychologist, 20, 535.

Lindsley, 0. R. (1956). Operant conditioningmethods applied to research in chronic schizo-phrenia. Psychiatric Research Reports, 5,118-139.

Lindsley, 0. R. (1957). Conditioned suppres-sion of behavior in the dog and some sodiumpentobarbital effects. Unpublished doctoraldissertation, Harvard University.

Lindsley, 0. R. (1960). Characteristics of thebehavior of chronic psychotics as revealed byfree-operant conditioning methods. Diseasesof the Nervous System (Monograph Supple-ment 21), 66-78.

Lindsley, 0. R. (1963a). Experimental analysisof social reinforcement: Terms and methods.American Journal of Orthopsychiatry, 33,624-633.

Lindsley, 0. R. (1963b). Free-operant condi-tioning and psychotherapy. In J. H. Masser-man (Ed.), Current psychiatric therapies (Vol.3, pp. 47-56). New York: Grune and Stratton.

Lindsley, 0. R. (1966). An experiment withparents handling behavior at home. In J. J.Parnicky (Ed.), A Johnstone bulletin (Vol. 9,No. 1, pp. 27-36). Bordentown, NJ: EdwardR. Johnstone Training and Research Center.

Lindsley, 0. R. (1971). Precision teaching inperspective: An interview with Ogden R. Lin-dsley, Ann Duncan interviewer. Teaching Ex-ceptional Children, 3(3), 114-119.

Lindsley, 0. R. (1972). From Skinner to pre-

cision teaching: The child knows best. In J.B. Jordan & L. S. Robbins (Eds.), Let's trydoing something else kind of thing: Behavior-al principles and the exceptional child (pp. 1-11). Arlington, VA: Council for ExceptionalChildren.

Lindsley, 0. R. (1994). "Point/see/say" or"slide/see/say" learning channels. Journal ofPrecision Teaching, 12(1), 2-7.

Lindsley, 0. R. (1995). Ten products of fluency.Journal of Precision Teaching, 13(1), 2-1 1.

Lindsley, 0. R., Hobika, J. H., & Etsten, B. E.(1961). Operant behavior during anesthesiarecovery: A continuous and objective method.Anesthesiology, 22, 937-946.

Lovitt, T C. (1967). Free-operant preference forone of two stories: A methodological note.Journal of Educational Psychology, 58, 84-87.

Rayfield, F, Segal, M., & Goldiamond, I.(1982). Schedule induced defecation. Journalof the Experimental Analysis of Behavior, 38,19-34.

Reese, E. P (1966). The analysis ofhuman op-erant behavior. Dubuque, IA: William C.Brown.

Rosales-Ruiz, J., & Fitzimmons, P. A. (1996,May). The effects of response fluency on dis-crimination learning. Poster presented at themeeting of the Association for Behavior Anal-ysis, San Francisco.

Sidman, M. (1960). Tactics of scientific re-search. New York: Basic Books.

Silverman, K., Lindsley, 0. R., & Porter, K. L.(1991). Overt responding in computer-basedtraining. Current Psychology: Research andReviews, 9(4), 373-384.

Skinner, B. F (1938). The behavior of organ-isms. New York: Appleton-Century-Crofts.

Skinner, B. E (1958). Teaching machines. Sci-ence, 128, 969-977.

Skinner, B. F (1976). Farewell, my LOVELY!Journal of the Experimental Analysis of Be-havior, 25, 218.

Skinner, B. F (1983). A matter ofconsequences.New York: Knopf.

Whimbey, A. (1989). Analytical reading andreasoning. Stamford, CT: Innovative Sci-ences.

Whimbey, A. (1995). Analytical reading andreasoning (2nd ed.). Stamford, CT: InnovativeSciences.

ogden r. university behavior...

Documents