Sharing and Cooperation in Pigeons Introduction Table Of Contents History Title Page List of Figures, Tables Introduction Purpose, Method Results and Discussion General Discussion References Abstract

Introduction

This study was designed as an attempt to demonstrate sharing and cooperative-like responding in pigeons. Sharing is defined by Hake, Vukelich, and Olvera (1975) as (a) an increase in the correspondence in the numbers of reinforcers across trials between subjects, and (b) a method of responding within trials consisting of responses by only one subject which would produce the reinforcer for itself. Cooperation is define by Hake and Vukelich, similarly, as (a) an increase in the correspondence in the numbers of reinforcers across trials between subjects, and (b) a method of responding within trials consisting of responses by only one of the subjects which would produce the reinforcer for the other subject (vs. producing the reinforcer for the responding subject in sharing). In cooperation or sharing, the responding must be under the control of the cooperation or sharing procedure i.e., must be under the control of the programmed cooperation or sharing contingency) to be cooperative responding or share responding (Hake and Vukelich, 1972). Responding which resembles cooperative responding topographically, but which has not been shown to be under the control of the cooperation procedure will be referred to as "cooperative-like" responding (cf. Marcucella and Owens, 1975).

This experiment was conceived as part of a research program with the long range goal of developing methods for the study of complex social behaviors with non-human animals. A primary feature of this line of research is typical of operant research: the research progresses from simple to complex procedures and phenomena (i.e., from the thoroughly studied and less controversial phenomena and procedures to the less well studied and not-yet acknowledged phenomena and procedures). A second major feature of this research program is the requirement that in studying a social-behavioral phenomenon, two tests must be passed before a demonstration of the social behavior of interest may be accepted. First, the social-stimulus control of the behavior of one subject by the behavior of the other subject must be demonstrated. Second, control by the appropriate social contingencies (procedures) must be demonstrated.

The first step in studying any social behavior is to show that the behavior of interest is indeed social. Keller and Schoenfeld (1950, pp. 257-258, 352) define social behavior as behavior under the discriminative control of the behavior or products of behavior of another organism. Besides the logical appeal, for a behaviorist, of Keller and Schoenfeld's definition of social behavior, the definition is based on a traditionally accepted definition of social behavior (Keller and Schoenfeld, p. 352). Keller and Schoenfeld (and Allport, as cited by Keller and Schoenfeld) stress that social behavior and individual behavior are in no way different from one another, other than in the "origin" of the controlling stimuli. Social behavior only differs from individual behavior in that the stimuli controlling social behavior "originate" with another organism (generally accepted to be a conspecific, cf. Keller and Schoenfeld, pp. 257-258), rather than from a source unrelated to another organism.

The behavior of the stimulus organism controls the behavior of another organism via changes in the environment which accompany the "stimulus" behavior. For example, the behavior of an animal produces changes in the light which is being reflected from the stimulus animal to a subject animal. The behavior of the subject animal which is under the discriminative control of these particular changes in patterns of electromagnetic radiation (i.e., under the control of visual stimuli) is the social behavior. Other environmental changes which are produced by the behavior of organisms, and which serve as discriminative stimuli for the behavior of another organism, may not be as obviously of a social nature. That is, a social stimulus may include environmental changes which do not propagate in a "line of sight" manner from one animal to another (as in "sight" of the stimulus behavior), but whose effects propagate less directly and often with longer delays (e.g., a social stimulus such as a telephone bell which is operated by a person responding on a dial in another city). In addition, social stimuli may include events of a less immediately tangible nature, as in changes in response requirements produced for one animal by another (e.g., changes in response requirements produced during competitive or cooperative responding). Changes in response requirements as products of the behavior of one organism which control the responding of another organism (i.e., social stimuli) become especially important in the more complex social behaviors such as cooperation, sharing, competition, trust, etc.

It can be seen that be defining social behavior as behavior under the control of stimuli arising from the behavior of another organism (Keller and Schoenfeld, 1950, p. 257), any procedure in which the responding of one organism in any way and through any medium affects the responding of another organism is a social procedure, the affected behaviors are social behaviors, and the stimuli controlling the affected behaviors are social stimuli (i.e., a social "relation" exists, Hyten, 1983). Because we are ill equipped to deal with certain stimuli, and in order to produce interpretable results, some social stimuli will be eliminated as much as possible (e.g., tactile, auditory, and olfactory social stimuli) and others will be manipulated (e.g., sight of the partner subject, changes in reinforcement schedules produced by the responding of the other subject, and keylight changes produced by the other subject, as will be described later).

Given the multiplicity of possible inter-organism behavioral relationships inherently possible in a multiple organism situation, once a behavior is shown to be social the behavior still must be shown to be under the control of the social contingencies of interest. Such a demonstration is easiest in the "simpler" social behaviors such as matched-dependent behaviors. Social behaviors such as matched-dependent behaviors are simple in that the social stimuli controlling subject responding are generally the more directly and immediately observable environmental changes (e.g., the "sight" of the stimulus behavior). A matched-dependent behavior is one which occurs under the discriminative control of responding by the stimulus organism just prior to the responding of the subject animal and occurs in an environment which is similar to that of the subject animal. Matched-dependent responding has been studied by itself (Danson and Creed, 1970; Hake, Donaldson, and Hyten, 1983; Millard, 1979), as well as serving as the basic procedure for other studies. A matched-dependent procedure has been used as a basic component of a classroom demonstration of cooperative-like responding in pigeons (Skinner, 1953, Skinner 1962) as well as in a cooperation procedure using human children (Azrin and Lindsley, 1956), and served as the base procedure for this study. Social behaviors such as sharing, cooperation, and competition are more complex in that the stimuli controlling the responding may include changes in response requirements as produced by the responding of the other animal (and relatively long delays of reinforcement), as well as the more easily observed stimuli such as sight of the stimulus behavior.

The complexity of the stimulus control has been a major problem in demonstrating the types of social behaviors which this study attempted to produce. According to Marcucella and Owens (1975), "the occurrence of cooperative responding at least as defined by Skinner (1953) and Hake and Vukelich (1972) has yet to be demonstrated in a species as low on the phylogenetic scale as the albino rat." Hake and Vukelich state that in a social cooperation procedure the cooperative responding of one organism must be under the control of the responding of the other organism. Such control has not been shown in studies, to date, without contingencies in addition to the food-reinforcement contingencies (Daniel, 1942; Daniel, 1943), or with controls for the possibility of control by alternate contingencies such as a competitive contingency (Taylor and Espamer, 1971), as discussed below. The multiplicity of sources of control of cooperation (or cooperative-like) responding complicate any demonstration of cooperation. In a review of experiments reporting cooperative responding in non-human animals, Hake and Vukelich (1972) suggest sources of control in the experiments other than the cooperative responding of the paired animals or the reinforcers produced by the cooperative procedure.

For example, Daniel (1942) demonstrated a cooperative-like responding in rats. However, the contingencies maintaining the cooperative-like responding may have been a shock-avoidance contingency rather than the cooperative food reinforcement contingency. A food crock was placed on a shock grid with a switch platform at one end of the chamber (a rat could not feed and hold down the platform simultaneously). Shock was turned off as long as one rat was on the platform. The cooperative-like responding consisted of one rat staying on the platform (thereby turning off shock) while the other rate ate, and alternation of positions resulting in both rats receiving food. Daniel observed that the "rat on the platform would reach off, holding the platform down with only one foot, and nudge the feeding animal", as well as climbing on the back of the feeding rat or biting the tail of the feeding rat (p. 367). Since the responding of one rat affected the responding of the other rat, the procedure was social. However, the social behavior was not one of "allowing the other rate to eat", but one of avoiding socially signalled shock.

Control by the shock-avoidance schedule rather than the cooperative-food contingency is also probable in a second study by Daniel (1943) using the same procedure with the addition of a lid on the food crock. The lid on the food crock was open only as long as one rat was on the platform. The food crock was again placed on a shock grid. The procedure differed from that of Daniel, 1942, in that the probability of shock was varied from 100% down to 0% through the course of the study. As long as the probability of shock was above zero (i.e., as long as shock occasionally occurred when a rat left the platform) the rats alternated positions, resulting in the lid of the food crock being open as well as resulting in shock rarely occurring. When the probability of shock was reduced to zero, both rats stayed at the food crock. Since the food crock lid was only open while a rat was on the platform, the lid remained closed. Daniel continued the experiment until one rat of each pair starved, suggesting that the responding of the rats was not under the control of the food-reinforced cooperation procedure, but rather a socially signalled shock avoidance schedule.

Marcucella and Owens (1975) provided direct evidence for control by a signalled shock-avoidance contingency in the studies by Daniel (1942, 1943). With a pair of rats in a chamber similar to the one used by Daniel, the distance between the platform and the food cup was increased to a distance such that the rat on the platform was eventually unable to reach the rat at the food cup. Increasing the distance resulted in reduced alternation, suggesting control by the signalled avoidance procedure rather than the food-reinforced cooperation procedure. Therefore, cooperative feeding was not shown by Daniel (1942, Daniel 1943).

Hake and Vukelich (1972) point out similar problems in another study demonstrating cooperative-like responding in rats. In an experiment by Taylor and Espamer (1971) two rats were connected by a wire passed through the back of a two-compartment chamber such that only one rat could obtain food at a time. For one rat to reach the response bar, the other rat had to make the cooperative-like response (i.e., had to move to the back of its compartment). Hake and Vukelich suggest that not only was it questionable as to what was serving as a social stimulus, but it was also not clear what responses were under social control. That is, was movement of the partner rat toward the response bar controlling the responding of the other rat in its moving to the back of its compartment, or was the movement to the back of the compartment under the control of other stimuli such as the digging of the wire harness into the skin of a subject as the paired rat pulled toward the response bar? If the movement of one subject toward the response bar, per se, served as a social stimulus for the movement to the rear of the compartment by the paired subject, then the responding of the subject moving to the rear of the compartment was under the control of the cooperation procedure. However, if the cooperative-like responding of subjects was under the aversive control of the harness, then the responding was under the control of a social-avoidance procedure in which a subject was causing the harness to dig into the skin of the other subject. Control by a competitive procedure was also a possibility in that the rat pulling harder and for longer against the pull of the paired rat produced reinforcements for itself.

Boren (1966) produced cooperative-like responding in monkeys using a procedure in which subjects alternately produced food for each other. This procedure failed to maintain responding, a failure which suggests a lack of control by the appropriate social stimuli and procedures. Four monkeys were pre-trained individually on delay of reinforcement schedules. Two of the four were then trained as a pair on a forced alternation procedure in which one subject was required to respond on a chain schedule with a ratio component (with a stimulus light of a particular color presented) and a differential reinforcement of other-behavior (DRO) component (with a stimulus light of a different color presented). The ratio component produced food for the other monkey as well as producing the DRO component. The length of the DRO component was determined by the partner monkey (i.e., one monkeys DRO component corresponded to the others ratio component). All four monkeys were run as pairs in a "free responding" condition in which subjects could respond at any time to operate the food dispenser of the other monkey. Sessions were continued until a set number of food pellets had been delivered (140 pellets total for one pair and 70 pellets for either subject in the other pair). Responding decreased to such low levels that the required number of pellets were not being delivered in 24 hour sessions and "the likely final consequence was that both animals would eventually starve to death" (p. 696).

In the pair which was pre-trained on a forced alternation procedure, one subject often received a large proportion of its food reinforcement when not responding, while the other subject responded at high rates and received few reinforcements. Thus, one subject was receiving response independent reinforcement while the other was responding on a large variable-ratio schedule. The responding of the higher rate subject was characterized by response runs which terminated with the click of the food dispenser in the chamber of the other subject. This pattern indicates control of the responding of the high-rate subject by the feeder click of the other animal, but not necessarily the cooperation contingency (i.e., not by the cooperative contingency for responding only after having received a food reinforcement) programmed in the free responding condition. Subjects had been trained to "tolerate the delays of reinforcement likely to occur when the two monkeys were placed together" (Boren, 1966). The monkeys that received the forced alternation procedure learned to respond during one key color and not during another. None of the subjects were trained to respond differentially to the behavior of the other subject.

In a classroom demonstration of cooperative-like responding by pigeons, the behavior of subjects did appear to be under the control of the behavior of paired subjects (Skinner, 1962). Two pigeons were separated by a glass partition. On each side of the glass partition were three response keys arranged vertically over a food magazine opening. On each trial corresponding keys in each pigeons compartment were effective (one key in each compartment). Pecks on the effective key by both pigeons in their respective compartments at the same time produced food reinforcement for both pigeons. The operative pair of keys was randomly selected on each trial by the programming equipment. Non-social discriminative stimuli were not programmed. The subjects found the effective keys by responding on all the keys until the food magazine operated. The subjects pecked keys "in a perfect mirror image pattern under the control of each others behavior" (p. 533). Although no data was collected, the demonstration has been important in that the procedure has served as the basis for other social-behavior research, including this study.

The procedure upon which the current study was most directly based is the matched-dependent procedure of Hake, Donaldson, and Hyten (1983). In Hake, Donaldson, and Hyten, two rats were separated by a clear Plexiglas partition. On each side of the partition were two response keys arranged vertically over a food cup. On each trial the keys of the stimulus rat (the "leader" rat) were lighted. The first response by the leader on the randomly selected operative (correct) key lighted the response keys of the subject (the "follower" rat) and started the matching-to-sample period. Responding by the follower rat on the key corresponding to the one on which the leader was responding produced food at the end of the trial for the follower rat. Responding on the "correct" key by the leader during the matching interval produced food for the leader at the end of the trial.