Emily was a 12-year-old female with a limited vocabulary and little spontaneous speech ..... Smith, 1993) found that structured play environments produced more.
Journal of Autism and Developmental Disorders, Vol. 27, No. 2, 1997
Teaching Deception Skills in a Game-Play Context to Three Adolescents with Autism Dana R. Reinecke,' Bobby Newman, Anya L. Kurtz, Carolyn S. Ryan, and Nancy S. Hemmes Queens College and the Graduate Center/CUNY
Baron-Cohen (1992) found that students with autism are impaired in their ability to deceive. A multiple-baseline across-subjects design was conceptualized to test the hypothesis that such students could be taught to deceive. Two conditions were presented in baseline and treatment phases. In Condition 1, the student guessed in which hand a small object was hidden when the experimenter presented two closed fists. In Condition 2, the student hid the object and presented two closed fists to the experimenter for a guess. Reinforcement was delivered contingently upon independent guessing during Condition 1 in both baseline and treatment phases. Under Condition 2, reinforcement was delivered noncontingently during the baseline phase and contingently upon successive approximations to the target behavior of deception during the treatment phase. All students displayed the acquisition of at least three of the responses included in the deception response during the baseline phase, and two students showed an erratic acquisition of the total skill during the baseline phase. Results indicate that students with autism can leant to deceive, even without formal intensive training.
Sodian and Frith (as cited by Baron-Cohen, 1992), stated that individuals with autism are unable to manipulate another person's thoughts; that is, they are impaired in their ability to deceive. They postulated that this impairment is due to a cognitive deficit in theory of mind. Based upon their performance in a penny-hiding game, as compared to controls of normally functioning and mentally retarded peers, Baron-Cohen concluded that his autistic subjects were unable to engage in deceptive behavior. 'Address all correspondence to Dana Reinecke, Psychology Department, Queens College, Flushing, New York 11693. 127 0162-3257/97/0400-0127$ 12.50/0 C 1997 Plenum Publishing Corporation
128
Reinecke, Newman, Kurtz, Ryan, and Hemmes
Baron-Cohen {1992} stated that his autistic subjects were able to enjoy the game at the level of object occlusion (i.e., were able to hide the penny) but were unable to engage in information occlusion, defined as preventing the opponent from obtaining clues regarding the whereabouts of the penny. The subjects with autism were considered oblivious to the state of mind of the opponent, unable to recognize that others have knowledge and beliefs apart from their own. Unfortunately, the reasons cited for the apparent deficit provide no hope for overcoming this disability. We feel that Baron-Cohen's study attempted to measure deception as a play skill, providing insight into a real lack of skills in many individuals with autism. Most children with autism do not play appropriately, and defining their specific play deficits can help us to help them develop more normally. The present study hopes to show that deficits in deceptive play skills are not general to all persons with autism, and that they can be overcome in those in whom it is apparent.
METHOD Subjects and Setting
Three students diagnosed with autism by outside agencies served as subjects. All students functioned in the moderate to severe range of mental retardation. Emily was a 12-year-old female with a limited vocabulary and little spontaneous speech. Walt was a 14-year-old male with similar verbal capabilities. Eliot was a 9-year-old male with no verbal communication abilities and limited sign language. Based upon the observations of the experimenters throughout 9 months of intensive work with the students, it was judged that none showed appropriate play skills. During free time, all invariably engaged in perseverative behavior. As demonstrated by the pretest described later, all students were able to follow simple instructions. The study was conducted at an after-school program for children and adolescents with autism. All three students had been enrolled in the program for at least 9 months. The experimenters had also been employed as therapists at this program for at least those 9 months. All phases of the study, excluding the pretest, were conducted in a small lounge containing a large rectangular table and several chairs. The students never entered this room for any purpose other than to participate in this study, and two or more students were never present in this room at the same time. The game was played across the table. The student and the experimenter who served as his/her partner (hereafter called the partner) sat op-
Teaching Deception
129
posite from the experimenter who served as the opponent (hereafter called the experimenter). Design and Procedure Pretest
A pretest was conducted to demonstrate that the students were capable of following instructions. The pretest was conducted in the students' regular classroom. Discrete trial training was used to teach three instructions: "Pick it up," "Give it to me," and "Make a fist." A response was considered independent and correct if the student performed the action that was described without physical prompting. During the pretest period, sessions consisting of 60 trials each were conducted once per day. The responses to the three instructions were each trained in isolation for 10 trials and were then randomized throughout the remaining 30 trials. Reinforcement was delivered continuously upon independent correct responding for the first 30 trials and intermittently on a variable ratio-three schedule (VR 3) for the last 30 trials. Physical prompting was used when students failed to respond correctly to the first presentation of an instruction. Trials during which a second verbal or a physical prompt was required were not followed by reinforcement. Response Definitions Guessing, the response measured in Condition 1, was defined as pointing to or touching one of the experimenter's closed fists. If the student did not attempt to guess, the partner first verbally and then, if necessary, physically prompted the response. Deception, the behavior measured in Condition 2, was defined as engaging in five separate, operationally defined responses in a single trial of hiding the object (Baron-Cohen, 1992). (See Table I.) If the student failed to engage in any of these responses, the correct response was verbally and/or physically prompted. Baseline and Treatment The remainder of the experiment was planned as a multiple-baseline across-subjects design. Sessions lasted approximately 15 to 20 minutes each, once per day, for each student. All students began the baseline phase on the same day. Treatment was introduced to Emily on Day 26. Due to an
130
Reinecke, Newman, Kurtz, Ryan, and Hemmes
Table I. Five Operationally Defined Responses Included in the Total Target Response of Deception Response Object occlusion Hidden transfer Empty fist closed Hiding fist closed Not indicating
Definition Hiding the object in one fist so that it is not visible Hiding the hands under the table while engaging in object occlusion Keeping the fist not containing the object closed from the time the fists are presented until the exerimenter has guessed Keeping the fist containing the object closed fi-om the time the fists are presented until the experimenter has guessed Not indicating in which hand the object is hidden by verbally telling or shaking the fist
unexpected increase in demonstrated deception skills, treatment was not introduced with either of the other two students. Two conditions were present in all sessions across both baseline and treatment phases. Condition 1 was termed "student as guesser" and Condition 2 was termed "student as hider." Condition 1 always preceded Condition 2. Condition 1 consisted of the experimenter and partner first modeling two trials of the game, with the partner in the guessing role. The experimenter hid a small ball under the table in one hand and invited the partner to guess with the question, "Which hand is it in?" The experimenter then hid the ball and invited the student to guess for five trials. To provide motivation, the experimenter presented her fists with a ball in each fist for the first, third, and fifth trials of this phase. In this way, the student could not lose in at least three of tbe five trials. The ball was hidden in the left hand for the second trial and in the right hand for the fourth trial. If the student won, that is, chose the hand witb the ball in it by pointing to or touching it, s/he was congratulated for wintiing and given tbe opportunity to play witb the ball for a few seconds. If the student did not win, the experimenter played witb the ball and the student sat quietly. The partner recorded the data for the trial during the play period. Edible and social reinforcement was provided during Condition 1 in both basehne and treatment phases for independent guessing. If the student did not attempt to guess, the partner first verbally and then, if necessary, physically prompted the response. If a prompt was required, no reinforcement was provided. Condition 2 also began with the experimenter and tbe partner modeling two trials of the game. In tbis condition, however, the partner modeled the role of the hider. For Emily and Eliot, both of whom had a tendency to throw the ball, tbe hidden object was a poker chip that could be traded in for the ball. For Walt, the ball remained as the hidden object.
Teaching Deception
131
The chip or ball was placed in front of the partner, who hid it witb her hands under the table. She then presented her fists to the experimenter, who made a guess. The winner played with tbe ball for a few seconds. Tbe chip or ball was then placed iti front of the student and the partner instructed the student to "Hide the (chip/ball)." The experimenter attempted to guess incorrectly about half the time, so that the student won in 5 trials out of 10 to create a motivating and challenging condition. The winner of each trial was given the opportunity to play with the ball for a few seconds while the partner recorded data. Social reinforcement was delivered throughout botb baseline and treatment phases of Condition 2. Primary reinforcement, consisting of 10 small pieces of some desirable edible, was presented noncontingently at tbe end of each session during the baseline phase. During treatment, however, these primary reinforcers were delivered only following independent approximations to the desired target response of deception.
RESULTS
The data collected from the pretest demonstrated tbat all students were capable of instruction-following (Figure 1). Independent correct responding ranged from 80-100% for Emily and from 65-100% for Eliot. Independent correct responding for Walt ranged from 96-100%, with the majority of sessions scored at 100% {see Figure 1). While the data collected from Ehot were somewhat erratic, the lowest scores occurred within the first three sessions. Higher scores were evident during the last few sessions, indicating that the student needed to learn the particular responses demanded in the instruction-following demonstration before be was able to follow the instructions. Emily and Eliot both guessed reliably by Session 13. Independent guessing ranged from 0-80% for Emily from Sessions 1 through U, and remained at 100% thereafter, falling only twice (60% during Session 15, and 80% during Session 24) (Figure 2). For Elliot, percentages of independent guessing per session ranged from 0-100% before stabilizing at 100% from Session 13 on, falling once to 80% during Session 26. Walt's guessing remained erratic throughout the study. Percentages of independent guessing ranged from 20-100%. The greatest degree of stability can be seen in the last 12 sessions, where percentages of independent guessing per session remained at 100%, except for during Session 34, when indepetident respotiditig fell to 80%. Independent guessing remained at 100% for the final eight sessions (see Figure 2).
U2
Reinecke, Newman, Kurtz, Ryan, and Hemmes 100 90 80 70 60 50 40 50 20 10
-, -
Emily
0
100 90 -J 80 70 60 50 40 30 20 10 0
100 90 80 70 60 50 40 30 20 10
-| H H 4
0 0
Fig. 1. Demonstration of percentage of instruction-foUowing in three adolescents with autism.
The mean number of deception responses independently emitted per trial for each session was calculated and plotted on Figure 3. The mean number of deception responses independently emitted by Emily was erratic until treatment was introduced at Session 26. She never independently achieved a mean of more than three deception responses until Session 36, when active teaching methods had been implemented for 10 sessions. From this point, there is a clear upward trend in independent responding. The mean number of deception responses independently emitted by Ehot rose steadily until Session 18, at which point responding stabilized at
Teaching Deception
133 100 -| 90 80 70 •
60 50 40 30 20 10 0
-
Emily
0
100 90 80 70 60 50 40 30 20 10 0
10 15 20 25 30 35 40 45 50
i. Eliot 0
100 90 80 70 60 50 40 30 20 10 0
5
5 10 15 20 25 30 35 40 45 50
-| —T
u
1
1
1
\
10 15 20 ?e. M
1
1
1
\
35 40 45 50
Fig. 2. Percentage of independent guessing responses per session in three adolescents with autism.
a mean of three independent responses per trial. The mean number of independent responses ranged from 2.5-4.8 for the remainder of the study, with most of the sessions averaging three independent responses per trial. Responding reached a high of 4.8 during Session 43, after which it dropped back into the previous range. The data collected from Walt are erratic, but show an upward trend. The mean number of independent responses per trial ranged from 1.1-4.8. Scores as low as 1.3 were evident even at Session 30, immediately following a session in which the mean number of independent responses per trial was 4.5. All sessions in which the mean number of independent responses
Reinecke, Newman, Kurtz, Ryan, and Hemmes
134 5 -.
Baseline
4 -
'
Treatment
J'
T _
2 1 -
Emily 0
0
5
10 15 20 25 30 35 40 45 50
4 -
^^.S-..--V^V I-
3 2 1
0
5
10 15 20 25 30 35 40 45 50
5
10 15 20 25 30 35 40 45 50
5 4 3 -
1 0 0
Sessions
Fig. 3. Mean number of independent responses per trial included in the definition of deception.
per trial was 4 or higher occurred after Session 23, indicating some increasing trend. This phenomenon is more clearly evident in Figure 4, where the percentage of independent total deception responses (i.e., all five responses were achieved independently within one trial) for each session is plotted. Emily did not independently engage in the total deception response until Session 39 (13 sessions after treatment was implemented), when she independently displayed the response over 30% of the trials. Independent responding ranged from 0-40% in the remaining sessions.
Teaching Deception
135 '00 -| 90 80 70 60 50 AO 30 20 10 0 0
100 90 80 70 60 50 40 30 20 10 0
-| -
0
, Ehot
5
10 15 20 25 30 35 40 45 5D
Fig. 4. Percentage of independent total deception responses.
Eliot displayed the deception response during Session 42, over 40% of trials. During the next session, he responded accurately over 90% of the trials, before returning to his previous zero level. Walt independently displayed the deception response over 10% of trials at Session 12, and then failed to engage in the response again until Session 22. He then engaged in the response independently from 10-80% of trials for the remainder of the sessions, failing to respond at all in a session only twice. There is no regularity to the pattern of responding from Session 22 on, however.
136
Reinecke, Newman, Knrtz, Ryan, and Hemmes
DISCUSSION The goal of this study was to demonstrate the effectiveness of a particular method in teaching deception skills to students with autism. We were able to demonstrate acquisition of deception skills, observing effects even before the introduction of treatment for two students. These results indicate that although some students with autism may not display deception skills, as indicated by Baron-Cohen (1992), they may be able to learn to engage in deceptive behavior. Since treatment had not been introduced to Walt or Eliot, their acquisition of independent responding cannot be positively attributed to any independent variable. There are many possible reasons for Wah's and Eliot's acquisition of the deception skill prior to treatment. Strong social reinforcement may have contributed to the learning effects displayed by all students. In addition, the consequences for winning and the frequency with which it was assured that the students would win may have acted as intermittent reinforcement for correct playing. Also, repeated exposure to modeling by the experimenters, as well as practice effects, may have contributed to the learning process. We were able to demonstrate that our students are able to deceive in the context of the penny-hiding gme. We can therefore conclude that individuals with autism are capable of deception responses and that this ability can be fostered and developed in the proper environment. Deception skills are necessary in the competitive play of almost any game. It is as important to develop play skills in individuals with autism as it is to develop other functional daily living skills. Singh and Millichamp (1987) showed a correlation between improved social play skills and increased social interactions in their mentally retarded subjects. Foxx, McMorrow, and Schloss (1983) used a game to train social sldlls, finding that the improvements effected by their intervention were maintained in nontraining settings. In addition, much research has shown that the training of socially significant behavior, including playing, has been associated with decreases in maladaptive behavior and increases in other adaptive behavior (Singh & MUUchamp, 1987). One problem evident in many investigations of play skills with individuals with developmental disabilities is that only spontaneous play is examined. Although spontaneous play may certainly be a goal, it is premature to state that individuals with autism are unable to play without first attempting to elicit play, lingerer and Sigman (as cited by Jarrold, Boucher, & Smith, 1993) found that structured play environments produced more sophisticated and diverse play than was produced spontaneously in unstructured environments.
Teaching Deception
137
A typical objection to research that attempts to elicit play is that eliciting involves modeling, and therefore that the resultitig play is merely imitation (Jarrold et al., 1993). While this may be true, our goal was to teach deception skills in game play. Whether or not the students were imitating the experimenters, they developed a game-play skill that serves to improve social interactions by allowing them to compete successfully and fairly with others in game play situations. We can conclude that both previous and possible future failures to accomplish the goal of teaching play skills to individuals with autism are not due to inherent and irreversible deficits in the individuals, but to flaws in the teaching methods. We should therefore work to refine and improve our techniques, knowing that we are working towards an attainable and worthy goal.
REFERENCES Baron-Cohen, S. (1992). Out of sight or out of mind? Another look at deception in autism. Joumal of Child Psychology, Psychiatry, and Allied Disciplines, 33, 1141-1155. Foxx, R. M., McMorrow, M. J., & Schloss. C. N. (1983). Stacking the deck: Tfeaching social skills to retarded adults with a modified table game. Joumal of Applied Behavior Analysis, 16, 157-170. Jarrold, C , Boucher, J., & Smith, R (1993). Symbolic play in autism: A review. Joumal of Applied Behavior Analysis, 26, 281-305. Singh, N. N., & Millichamp, C. J. (1987). Independent and social play among profoundly mentally retarded adults: TVaining, maintenance, generalization, and long-term follow-up. Joumal of Applied Behavior Analysis, 20, 23-34.
