CHOICE BETWEEN SINGLE-RESPONSE AND MULTICHOICE TASKS IN HUMANS.
Many studies have investigated choice using concurrent-chain procedures (Duncan & Fantino, 1972; Fantino, 1969; Grace, 1996; Green & Rachlin, 1996; Siegel & Rachlin, 1995; Takahashi, 1993). In this procedure, a subject is generally presented with two response-keys in the initial links of the chain, each of which is associated with a variable-interval schedule. Meeting the schedule requirement on either key produces a terminal-link schedule that leads to a primary reinforcer. Preference for the terminal-link schedules is examined by measuring choice responses during the initial links.
Catania (1975) defined the availability of choice as an element of freedom. The availability of choice meant that a subject had an opportunity to choose among multiple alternatives. Then, Catania (1975) demonstrated that pigeons preferred the terminal link that had multiple alternatives to the one with only a single-response alternative. In his experiment, one terminal link had two keys such that the pigeons could choose between the keys, and the other terminal link had only one key. Although delivered reinforcement rates were identical in the two terminal links, the pigeons preferred the link with two keys. He then proposed that the pigeons preferred the opportunity to choose among alternatives.
This phenomenon has been reported not only in pigeons (Catania, 1980; Catania & Sagvolden, 1980; Cerutti & Catania, 1986), but also in rats (Voss & Homzie, 1970) and humans (Suzuki, 1997). These experiments used identical alternatives in both terminal links. That is, the previous studies have indicated that the subjects prefer the opportunity to choose among the alternatives when all of them are identical.
Some studies have reported that subjects do not always prefer the terminal link that has multiple alternatives when the alternatives differ (Hayes, Kapust, Leonard, & Rosenfarb, 1981; Leigland, 1979; Solnick, Kannenberg, Eckerman, & Waller, 1980; Suzuki, 1997, 1999). Especially Hayes et al. (1981) reported that pigeons preferred a single-response task over a multichoice task even though there was no difference in the reinforcement rate when the latter task included a less preferred alternative. Therefore, they suggested that the subjects did not always prefer the opportunity to choose among different alternatives when a less preferred alternative was one of them.
The present study examined choice between a single-response and a multichoice task in humans under two conditions. The present study had two purposes. The first purpose was to examine whether the degree of preference for the multichoice task was related to the number of alternatives in it. If the preference was due only to the opportunity to choose among identical alternatives, it seems reasonable to suppose that an increase in the number of alternatives should not change the degree of preference.
The second purpose was to examine the influence of the efficacy of choice on preference for the multichoice task. The efficacy of choice indicates the acquisition of the outcomes might depend on the choice among the alternatives. In other words, it means that the choice of the one alternative might produce the outcome and the choice of the other one might not. The alternatives used in the previous studies, which reported preference for the multichoice task, always produced identical reinforcers. This means that choice among the alternatives had no influence on the acquisition of the reinforcers, that is, all choices had the same outcome meaning that they were equally effective in producing the same outcome. The question is whether the preference for the multichoice task depended on the efficacy of choice other than the opportunity for choice among the alternatives.
To manipulate efficacy of choice, some subjects were instructed that the alternatives all had the same outcomes. This information manipulation was presented in order to tell the subjects that acquisition of the points was determined prior to their selection of an alternative. The other subjects were instructed that the alternatives were arranged to provide different outcomes. This information manipulation was presented in order to tell them that acquisition of the points might depend on which alternative they selected. It was predicted that the subjects in the former group would reduce their preference if the preference for the multichoice task came about because the alternatives produced different outcomes.
Fifty-three undergraduate students recruited from a course in introductory psychology at Hokkaido University participated in the experiment. Their ages ranged from 18 to 21 years. They were divided into two groups: the two-choice group (14 male, 14 female) and the five-choice group (13 male, 12 female).The data from one male of the two-choice group and two male and one female of the five-choice group were excluded from analysis because of computer problems.
Experimental sessions were conducted in three small rooms that each contained a personal computer (NEC PC9800-RX, VX, and XL), a desk, and a chair. Thus, 3 subjects could participate in the experiment at the same time. The response keys used in the experiment were located at the right-hand corner of the keyboard and were marked by putting letter signs on them. In the two-choice group, four keys were marked: I, II A, and B. In the five-choice group, seven keys were marked: I, II, A, B, C, D, and E.
The present experiment consisted of two conditions: the no-information condition and the information condition.
No-information condition. In this condition, the two-choice group had a choice between two tasks, one with only one response (i.e., the single-response task) and the other with two alternatives (i.e., the two-choice task). Figure 1 depicts a schematic diagram of the experiment.
In the initial link, green stimulus cards for each of the tasks were shown on the screen. The stimulus cards for the two tasks were separated by a thin line in the center. A press on the I-key produced the terminal-link task on the left, and a press on the II-key produced the terminal-link task on the right. The stimulus card positions varied randomly among the subjects.
The subject could obtain 10 points with a probability of .4 when choosing a card in the terminal link. The number of points available and the probabilities of obtaining them were shown on the upper part of the screen throughout a trial. Therefore, the subjects could always look at them during the sessions.
A single press on the I-key or the II-key was followed by a 0.5-second blackout. Next, the chosen stimulus cards appeared on the screen, and the terminal link started.
When the subjects chose the two-choice task, two rectangular green stimulus cards (2.5 cm x 5 cm) appeared on the screen. A press on the A-key chose the card placed on the left, and a press on the B-key chose the card placed on the right.
In contrast, when the subjects chose the single-response task, a single stimulus card appeared on the right or the left. The card position varied randomly from trial to trial. A press on the A-key was the response when the card was at the left position, and a press on the B-key was the response when the card was at the right position. However, a press on a key without its stimulus card present produced no programmed consequences.
In either task, after a response, the cards disappeared from the screen and the screen displayed "10 points" when the subject obtained 10 points, or "0 points" when no points were obtained. The points were presented for 5 seconds. However, the cumulative number of points was not revealed to the subjects. After this outcome phase, the initial link started again. When 40 trials were completed, this condition ended.
The points were exchangeable for money with 3 yen (equal to about 2.3 cents) per 10 points after the experiment. There were no differences in the probability of earning the points per response between the single-response and the two-choice tasks.
The five-choice group had a choice between the single-response and a five-choice task. The procedure was the same as that of the two-choice group except for using a five-choice task instead of the two-choice task. The A-key, the B-key, the C-key, the D-key, and the E-key corresponded to the cards, which were arranged from left to right on the monitor. And in the single-response task, a card appeared randomly in one of five positions.
The subjects were individually asked to enter the experimental room and sit in the chair facing the computer. They were given the following instructions in Japanese:
Please read this carefully. If it is difficult for you to understand this, do not hesitate to ask questions. However, there is some information that you may not get, because of the nature of the experiment.
In this experiment, your purpose is to earn as many points as you can, by choosing a card in a group. The groups or the cards are shown on the screen of the personal computer. When you make a choice of a group or a card, press one of the keys on the right-hand side of the keyboard, onetime. The keys are marked to indicate each group or each card. After choosing a card, you can get some points depending on your choice.
During this task, you must first choose one of the groups. And then, you must choose one card from the group which you have selected. The cards in the two groups are displayed on the screen, so that you can make a choice between the groups.
The number of points and the probability of getting points for each card are always displayed on the upper part of the screen as in the following example,
ex. Green card : 40% chance to win 10 points
This means that when you choose the green card, you can get 10 points with a probability of 0.4 and 0 points with a probability of 0.6. Please pay attention to the number of points and the probability of earning points, before making a choice.
Notice that the cards are independent of each other. Even if two identical colored cards appear, there is no relationship among them. In addition, keep in mind that new cards appear on each trial.
After choosing one card, the number of points given to you is displayed on the screen. You are told only the number of points for each choice. After the experiment, the points you have earned will be exchanged for money at a rate of 3 yen per 10 points.
Please call the experimenter, when the message "The experiment has finished!!" appears on the screen.
After reading this instruction, the subjects had a set of training trials. The procedure of the training trials was the same as in the main experiment. The training trials continued until the subject verbally answered that he or she understood the procedure. After finishing the training trials, the experimenter went out of the room and the session started.
Information condition. The subjects were studied in the information condition immediately following the no-information condition. The two-choice group was divided into two subgroups: The two-choice/synchronism group (n = 14) and the two-choice/asynchronism group (n = 13). Also, the five-choice group was divided into two subgroups: The five-choice/synchronism group (n = 10) and the five-choice/asynchronism group (n = 12).
The procedure in the information condition was identical to that in the no-information condition except that some information was presented to the subject after the display of the number of points earned on a given trial. This information told the subjects the number of alternatives associated with points (i.e., "lucky cards") in the trial just finished. The number of lucky cards was indicated for each task. The information was displayed for 3 seconds. After the information, the initial link started again.
The two-choice/synchronism group was presented with information indicating that the number of lucky cards was two or zero in the two-choice task. Also, the five-choice/synchronism group was presented with information indicating that the number of lucky cards was five or zero in the five-choice task. This information was presented in order to tell the subjects that the cards did not differ from each other so that their choice among them had no influence on the outcome.
In contrast, the two-choice/asynchronism group was presented with information indicating that the number of lucky cards varied randomly from two to zero in the two-choice task according to a quasi-normal distribution. Also, the five-choice/asynchronism group was presented with information indicating that the number of lucky cards varied randomly from five to zero in the five-choice task. For example, a subject may have one lucky card in one trial and four lucky cards in another trial. This information was presented in order to tell the subjects that the acquisition of the points might depend on their choice among the alternatives.
Of course, in the single-response task, the number of lucky cards was one or zero for all of the four subgroups. However, the delivery of points was independent of the number of lucky cards indicated by the information in any task. The outcome sequence of getting points in the single-response task was identical to those of the two-choice or the five-choice tasks in both the synchronism and the asynchronism groups. There was no difference in the outcome sequence of getting points between the two-choice and the five-choice group and between the information and the no-information conditions.
Before the session started, subjects were provided with the following instructions about the display of information.
Now, the next experiment starts. The procedure and your behavior are the same as in the previous experiment except for the following.
After the number of points is displayed, we tell you the number of lucky cards in the choice which you have just made. For example, "The Number of Lucky Cards in Group-I was 0" indicates that the group chosen by pressing the I-key had no lucky cards. And "The Number of Lucky Cards in Group-II was 3" indicates that the group chosen by pressing the II-key had three lucky cards. Of course, we tell you about the lucky cards for both tasks. You can utilize this information as you wish. Please earn as many points as you can in this experiment too.
After the subjects had read this instruction, they had training on at least two trials. The procedures were identical to those in the main experiment. The session started after the training trials.
The following analysis was conducted based on the data after the 11th trial to ensure that the subjects had some experience with the task and the response outcomes.
Choice proportions for the multichoice tasks of the two- and five-choice groups were calculated by dividing the number of choices of the multichoice tasks in the initial links by the number of trials after the 11th trial (i.e., 30). The left column of Table 1 shows the choice proportions of the two groups in the no-information condition.
One sample t test indicated that choice proportions for the multichoice tasks were significantly greater than 0.5 both in the two-choice group, t(26) = 8.62, p [less than] .0001, and in the five-choice group, t(21) = 5.70, p [less than] .0001. A Welch's t test revealed that the choice proportion for the five-choice group was significantly larger than that for the two-choice group, t(27.1) = 14.4, p [less than] .0001.
In the information condition, the two- and five-choice group were each divided into two subgroups. There were no significant differences in choice proportions in the no-information condition between the subgroups in the two-choice group, t(25) = .57, p [less than] .6, or in the five-choice group, t(20) = .03, p [less than] .98.
Table 1 shows the choice proportions of each subgroup in the no-information and the information conditions. Choice proportions in the information condition were significantly greater than .5 for all four subgroups; 1(13) = 4.5, p [less than] .0005 in the two-choice/synchronism group; t(12) = 2.6, p [less than] .05 in the two-choice/asynchronism group; t(9) = 2.9, p [less than] .05 in the five-choice/synchronism group; t(11) = 3.6, p [less than] .01 in the five-choice/asynchronism group.
Comparisons of each of the subgroups between their choice proportions during the no-information and the information conditions were conducted in order to examine the effects of information. In the two-choice group, there was no significant difference between the choice proportions for the two conditions; t(13) = .64, p [less than] .6 in the two-choice/synchronism group; 1(12) = .61, p [less than] .6 in the two-choice/asynchronism group. Similarly there was no significant difference in the five-choice/asynchronism group, t(11) = 1.04, p [less than] .4. In the five-choice/synchronism group, there was a tendency for a decrease of the choice proportion in the information condition compared with the no-information condition, but the difference was not significant, t(9) = 1.9, p [less than] .09.
The purpose of the information condition was to examine whether the information decreased the preference for the multichoice task. It is supposed that the information had influence on the subjects who had a choice proportion of more than .5 in the no-information condition. Figure 2 shows the choice proportion for each subject of the four subgroups under the no-information and the information conditions.
The number of subjects, who chose the multichoice task more than .5, was 13 in the two-choice/synchronism group, 13 in the two-choice/asynchronism group, 11 in the five-choice/asynchronism group, and 8 in the five-choice/synchronism group. The choice proportion of the five-choice/synchronism group in the information condition was significantly smaller than that in the no-information condition, t(7) = 3.3, p [less than] .05. However, in the other three subgroups, there were no significant differences (the two-choice/synchronism group, t(12) = 1.63, p [greater than] .13; the two-choice/asynchronism group, t(12) = .61, p [greater than] .55; the five-choice/asynchronism group, t(10) = .978, p [greater than] .35).
From the present experiment, the following results were obtained: First, the degree of preference for the multichoice task depended on the number of alternatives in it. Second, information that reduced the efficacy of the choice among the alternatives decreased the preference of the five-choice group. However, the subjects still preferred the multichoice task even after they had been exposed to such information.
The present experiment indicated that the preference for the multichoice task depended on the number of choice alternatives. In contrast, Catania (1980) reported that pigeon's preference for a multichoice task was independent of the number of alternatives. Besides the obvious species difference between these studies, there is an important procedural difference as well.
Whereas Catania (1980) used certain alternatives that always produced a reinforcer, the present experiment used probabilistic alternatives that only occasionally produced a reinforcer. Therefore, the opportunity of choice could be distinguished from the efficacy of choice in the present experiment, but not in Catania (1980). Because the acquisition of the outcome might vary depending on the choices in the present experiment, but might not vary in Catania (1980). It is possible that the phenomenon that the degree of the preference for the multichoice depended on the number of alternatives was caused by the use of the probabilistic alternatives and the perception of the efficacy of choice. But, the present study could not examine this hypothesis directly. Further study should explain why the degree of the preference for the multichoice depended on the number of alternatives.
In the present study, the result of the information condition indicated the possibility that the information which reduced the efficacy of choice decreased the preference for the multichoice task. This information told the subjects that the cards did not differ from each other. Therefore, they perceived, it was assumed, that their choices among them had no influence on the outcome.
Some studies reported that perception of contingency between choice and outcome had an influence on human behavior. Langer (1975) made her subjects choose one of lotteries. As a result, she reported that they evaluated their lottery highly when they could choose it by themselves more than when they could not. She described this phenomenon as an illusion of control, which was defined as an expectancy of a personal success probability inappropriately higher than an objective probability would warrant. In other words, she proposed that the subjects felt there was a higher probability of drawing a winning ticket in a lottery when they could choose by themselves.
In the present experiment, the multichoice task had some alternatives, and the subjects could choose one of them. Also, the subjects received the instruction which emphasized that the alternatives were independent of each other. Therefore, it is likely that these features of the present experiment induced the subjects to perceive that they could draw the "hit" alternative. On balance, it is likely that the information presented to the synchronism group under the information condition reduced such illusion so that the preference for the multichoice task decreased. Because the information meant that the alternatives were equally effective in producing the same outcome.
In the present experiment, the information which reduced the efficacy of choice decreased the preference for the multichoice task in the five-alternative group, but not in the two-alternative group. The two groups differed only in the number of alternatives in the multichoice tasks. However, the present experiment could not make clear how the number of alternatives and the efficacy of choice contributed to the preference for the multichoice task. Further study should examine this unsettled question.
In summary, the present study examined choice in human subjects between single-response and multichoice tasks. The result of the present experiment indicated that the degree of preference for the multichoice task increased as the number of alternatives increased. Also, the reduction of the efficacy of choice among alternatives decreased the preference for the multichoice task, but the degree of the preference still remained significant. The present study revealed that the preference for the multichoice task depended on the number of alternatives and the efficacy of choice among them.
This article is based on a dissertation submitted to Hokkaido University in partial fulfillment of the requirements for the doctoral dissertation.
My special thanks are due to I. H. Iversen, T. Takigawa, and M. Takahashi for reading the manuscript and making a number of helpful suggestions.
Corresponding concerning this article should be addressed to Shuji Suzuki, Department of Behavioral and Brain Sciences, Primate Research Institute, Kyoto University, Inuyama, Aichi 484, Japan.
CATANIA, A. C. (1975). Freedom and knowledge: An experimental analysis of preference in pigeons. Journal of the Experimental Analysis of Behavior, 24, 89-106.
CATANIA, A. C. (1980). Freedom of choice: A behavioral analysis. The Psychology of Learning and Motivation, 14, 97-145.
CATANIA, A. C., & SAGVOLDEN, T. (1980). Preference for free choice over forced choice in pigeons. Journal of the Experimental Analysis of Behavior, 34, 77-86.
CERUTTI, D. T., & CATANIA, A. C. (1986). Rapid determinations of preference in multiple concurrent-chain schedules. Journal of the Experimental Analysis of Behavior, 46, 211-218.
DUNCAN, B., & FANTINO, E. (1972). The psychological distance to reward. Journal of the Experimental Analysis of Behavior, 18, 23-34.
FANTINO, E. (1969). Choice and rate of reinforcement. Journal of the Experimental Analysis of Behavior, 12, 723-730.
GRACE, R. C. (1996). Choice between fixed and variable delays to reinforcement in the adjusting-delay procedure and concurrent chains. Journal of Experimental Psychology: Animal Behavior Processes, 22, 362-383.
GREEN, L., & RACHLIN, H. (1996). Commitment using punishment. Journal of the Experimental Analysis of Behavior, 65, 593-601.
HAYES, S.C., KAPUST, J., LEONARD, S. R., & ROSENFARB, I. (1981). Escape from freedom: Choosing not to choose in pigeons. Journal of the Experimental Analysis of Behavior, 36 1-7.
LANGER, E. J. (1975). The illusion of control. Journal of Personality and Social Psychology, 32, 311-328.
LEIGLAND, S. M. (1979). Deviations from matching as a measure of preference for alternatives in pigeons. Journal of the Experimental Analysis of Behavior, 32, 1-13.
SIEGEL, E., & RACHLIN, H. (1995). Soft commitment: Self-control achieved by response persistence. Journal of the Experimental Analysis of Behavior, 64, 117-128.
SOLNICK, J. V., KANNENBERG, C. H., ECKERMAN, D. A., & WALLER, M. B.(1980). An experimental analysis of impulsivity and impulse control in humans. Learning and Motivation, 11, 61-77.
SUZUKI, S. (1997). Effects of number of alternatives on choice in humans. Behavioural Processes, 39, 205-214.
SUZUKI, S. (1999). Selection of forced- and free-choice by monkeys (Macaca fascicularis). Perceptual and Motor Skills, 88, 242-250.
TAKAHASHI, M. (1993). Psychological distance to reward in monkeys. Behavioural Processes, 30, 299-308.
VOSS, S. C., & HOMZIE, M. J. (1970). Choice as a value. Psychological. Report, 26, 912-914.
Choice Proportions for Multichoice Task for Each Group in Information and No-Information Conditions. Condition Group No-Information Information Two-Choice Synchronism .617 (.012) .593 (.006) Asynchronism .610 (.006) .590 (.016) Overall .614 (.009) Five-Choice Synchronism .707 (.034) .593 (.010) Asynchronism .697 (.034) .647 (.020) Overall .702 (.032)
Note. Standard deviations are in parentheses. Data are based on performance after 11th trial. In the no-information condition, the synchronism and asynchronism groups experienced the identical contingency. In the information condition, the procedures which they experienced differed only in the information about the "lucky" cards.
|Printer friendly Cite/link Email Feedback|
|Publication:||The Psychological Record|
|Date:||Jan 1, 2000|
|Previous Article:||PROBABILITY OF STIMULUS EQUIVALENCE AS A FUNCTION OF CLASS SIZE VS. NUMBER OF CLASSES.|
|Next Article:||REVERSAL LEARNING IN CONDITIONAL DISCRIMINATIONS IS NOT CONTROLLED BY REINFORCER DENSITY.|