Teaching serial position sequences to monkeys with a delayed matching-to-sample procedure

Comparison was made of two methods for training monkeys to “observe” a two-member serial position sequence by pressing two consecutively lighted keys and then to “report” the sequence by pressing the same two keys in the same order but without the lights. A fading technique involving gradual elimination of brightness cues from “reporting” keys was found more effective than a no-fading procedure in which the cues remained bright during training and then were suddenly removed. Animals that failed to learn to report a new sequence with the no-fading procedure sometimes developed behavior incompatible with that desired. They made repeated and specific errors that prematurely terminated trials of the sequence to-be-learned, even though the correct key was cued by a bright light. They behaved appropriately, however, on succeeding trials of other sequences. Thus, the errors were followed by trials on which reinforcement occurred. Manipulation of this contingency indicated its importance in maintaining the stereotyped error patterns.     

Some factors that influence the acquisition of complex, stereotyped, response sequences in pigeons

Two pigeons were required to peck six to nine illuminated response keys. A response on any one of the keys darkened that key. When each key had been darkened, a reinforcer was delivered. No specific sequence of key pecking was ever required. The keys were presented in various matrices: three by two, three by three, horizontal rows, and vertical columns. The keys either presented the same stimulus, white light; or each key presented a different stimulus, a color or form. The results indicated that although there were 720 to 362,880 different sequences that would produce reinforcement, each bird developed a particular, stereotyped sequence that dominated its behavior. Variability among the birds across phases yielded less than 60 sequences, .0001 to 6 percent of the possible sequences. The data suggest that a reinforcement contingency that includes “free choice” of response sequence will produce stereotypical response sequences that function as complex “units” of behavior.     

Mirror pecking and timeout under a multiple fixed-ratio schedule of food delivery

Pigeons were trained to peck a key under a multiple fixed-ratio 25 fixed-ratio 175 schedule of food presentation. In the first condition, either a mirror or the opportunity to produce a 30-second timeout were available. In a second condition, mirror and timeout availability were reversed for the two groups. Following a return to the initial condition, mirror and timeout keys were presented together for all birds. Mirror and timeout responses occurred predominantly in the pause in the larger fixed-ratio component, regardless of whether the opportunities for the two responses were available singly or together. Mirror responding occurred in a greater proportion of the pauses than did timeouts. When the opportunities for both mirror pecking and timeout were available concurrently, they occurred with probabilities similar to those under the single conditions. Within the pause itself, mirror responses most frequently occurred immediately after reinforcement. Timeouts occurred most frequently toward the end of the pause, and some timeouts occurred in the early part of the run. Longer preratio pausing occurred in the larger fixed-ratio component in the conditions in which the mirror was present, whether or not any mirror pecks were recorded.     

The repeated acquisition of behavioral chains

Monkeys were trained with food reinforcement in a chamber containing four groups of three levers. For each session the monkey's task was to learn a new four-response chain by pressing the correct lever in each group. A stable pattern of learning resulted, and the number of errors reached a steady state from session to session. The technique was then used to determine how various durations of timeouts, following errors, affected the acquisition of new chains. With no timeout, the monkeys made a great many errors, due in large part to superstitious responses within the reinforced chain. Timeout durations ranging from 1 sec to 4 min reduced the number of errors substantially. A second experiment investigated the effects upon acquisition errors of presenting a single light (an “instruction” stimulus) over the correct lever. When this light did not influence the monkeys' responses to the three alternatives, the chains were learned as without it. When the light did control responding, the monkey pressed the appropriate sequence of levers but did not learn the sequence. Thus, when the light was removed, the monkey performed as if learning that sequence for the first time.     

Short-term memory in the rhesus monkey: A behavioral analysis of delayed-response performance

This study obtained quantitative data on the bodily orientations of rhesus monkeys in a delayed-response task and determined whether such orientations mediate the correct response in a choice trial. The basic task was a two-key chain schedule with the key leading to food signaled in the initial component. During the subsequent delay interval, the signal was removed, but it was necessary that one of the keys be pressed to advance the schedule to the terminal choice component. The position of the key pressed thus indicated orientation during the delay interval. When the monkeys had free access to the left and right keys, they tended to press the key leading to food throughout the chain schedule components and received food on more than 85% of the trials, even when the delay was extended to 20 seconds. However, when orientation toward the food key was disrupted by forcing the monkeys to press an extraneous center key during the delay, choice performance deteriorated. Requiring the center key presses early, rather than late, in the delay component had a strong disruptive effect. The relation of the results to the mediating coding-response hypothesis is discussed.     

Discrimination of brightness differences by rats with food or brain-stimulation reinforcement

Rats were trained to respond to the brighter of two keys. Four animals were trained with food pellets and four with electrical brain stimulation. Each discrimination sequence was initiated when the animal broke a light beam at the rear of the chamber, turning on the key lights and starting a 30-sec reinforcement period. An initial response on the brighter key was immediately reinforced, and further responses on the brighter key were then intermittently reinforced. Any time the dimmer key was pressed, a 30-sec timeout was introduced. During timeout, no response had any programmed consequence. When the reinforcement period or the timeout ended, a new discrimination sequence could be initiated. Daily 1-hr training sessions were conducted, and after seven or eight sessions, all animals were at or near errorless performance levels. The luminance of the brighter key was then systematically reduced, in seven steps, with two 30-min test sessions at each step. Orderly psychometric functions were generated for individual animals. Initial acquisition, once position preferences were broken, was equally rapid for food and for brain-stimulation animals, and the two reinforcement procedures yielded comparable levels of brightness discriminability.     

Several methods for teaching serial position sequences to monkeys

Three keys were available for monkeys to press. The objective was to teach the animals to press the keys in sequences up to 10 members in length. With fading procedures, a light that cued the correct key at a given serial member of the sequence faded out slightly each time the animal selected it, and became slightly brighter after the animal made an error at that sequence member. The correct keys were faded out, starting from the end of the sequence and proceeding toward the beginning. With control procedures, the cue lights were turned off suddenly, rather than being faded gradually. In almost every instance, the animals learned a longer series of unlighted key positions with the fading procedures than they did when each key-light was turned off suddenly. Also, requiring the animals to press a series of keys cued by lights before they could reach the sequence members they were to learn hampered them in learning the later serial members. By using several different sequences, it was possible to replicate these findings within the individual animal.     

Stimulus control of the pigeon's ability to peck a moving target.

Two pigeons were trained to peck whichever of eight keys displayed a white field (SD). The other seven keys displayed a white "X" on a black background (S delta). Each peck to SD produced three-second access to grain, a three-second intertrial interval (ITI), and the next trial. Pecks to S delta produced a three-second timeout (TO) and the same trial. During later sessions the key displaying SD changed every t seconds (t = 3, 2, 1, .5, and .25 sec), requiring the birds to track the position of the SD. Pecks on a ninth key increased t. Several sessions employed novel stimuli to ascertain the controlling stimulus dimensions. Both birds made few errors acquiring the original discrimination. During the tracking sessions, both birds made few errors when t = .5 sec. Only one reliably lengthened t. Data from sessions with novel stimuli indicate that color and form were important aspects of SD and S delta respectively; movement contributed to the final performance.     

Increasing the variability of response sequences in pigeons by adjusting the frequency of switching between two keys.

Three experiments compared the amounts of behavioral variability generated with two reinforcement rules. In Experiments 1 and 2 pigeons received food whenever they generated a sequence of eight pecks, distributed over two keys, provided that the sequence contained a certain number of change-overs between the keys. Although no variability was required-the birds could obtain all reinforcers by repeating the same sequence-the pigeons emitted a large number of different sequences. In Experiment 3 pigeons received food whenever they generated a sequence that had not occurred during the last 25 trials. After prolonged training, the birds showed more sequence variability than in the first two experiments. The analysis of the internal structure of the response sequences revealed that, in general, (a) the location of the first peck was highly stereotyped; (b) as the trial advanced, the probability of switching to the initially preferred key decreased whereas the probability of switching to the other key increased; and (c) a first-order Markov chain model with transition probabilities given by a logistic function accounted well for the internal structure of the birds' response sequences. These findings suggest that, to a large extent, the variability of response sequences is an indirect effect of adjustments in changeover frequency.     

Cognitive imitation in 2-year-old children (<Emphasis Type="Italic">Homo sapiens</Emphasis>): a comparison with rhesus monkeys (<Emphasis Type="Italic">Macaca mulatta</Emphasis>)

Here we compare the performance of 2-year-old human children with that of adult rhesus macaques on a cognitive imitation task. The task was to respond, in a particular order, to arbitrary sets of photographs that were presented simultaneously on a touch sensitive video monitor. Because the spatial position of list items was varied from trial to trial, subjects could not learn this task as a series of specific motor responses. On some lists, subjects with no knowledge of the ordinal position of the items were given the opportunity to learn the order of those items by observing an expert model. Children, like monkeys, learned new lists more rapidly in a social condition where they had the opportunity to observe an experienced model perform the list in question, than under a baseline condition in which they had to learn new lists entirely by trial and error. No differences were observed between the accuracy of each species' responses to individual items or in the frequencies with which they made different types of errors. These results provide clear evidence that monkeys and humans share the ability to imitate novel cognitive rules (cognitive imitation).     

Conjoint schedules of timeout deletion in pigeons.

This experiment attempted to bring behavior under joint control of two distinct contingencies, one that provided food and a second that extended the periods during which that food was available. Pigeons' responses on each of two keys were reinforced according to a single random-interval schedule of food presentation except during signaled timeout periods during which the schedule was temporarily disabled. By means of a conjoint schedule, responses on the initially less preferred key not only produced food but also canceled impending timeouts. When behavior came to predominate on this conjoint alternative, the consequences of responding on the two keys were reversed. Responding in 3 of 4 pigeons proved sensitive to the conjoint scheduled consequences, as evidenced by systematic shifts in response rates favoring the conjoint key. In 2 of these 3 pigeons, sensitivity to the conjoint contingency was evident under time-in:timeout ratios of 2:1 (time-in = 120 s, timeout = 60 s) and 1:5 (time-in = 30 s, timeout = 150 s), whereas for the other pigeon preference for the conjoint key was observed only under the latter sequence of conditions. There was only weak evidence of control by the conjoint scheduled consequences in the 4th subject, despite extended training and forced exposure to the conjoint alternative. The overall pattern of results is consistent with studies of timeout avoidance but also shares features in common with positively reinforced behavior.     

Perseverative errors in object discrimination learning by aged Japanese monkeys (Macaca fuscata).

To examine the nature of age-dependent cognitive decline, performance in terms of concurrent object discriminations was assessed in aged and nonaged Japanese monkeys (Macaca fuscata). Aged monkeys required more sessions and committed more errors than nonaged ones in the discriminations, even in simple object discriminations. Analyses of errors suggest that aged monkeys repeated the same errors and committed more errors when they chose a negative object at the 1st trial. A hypothesis analysis of behavior suggests that their incorrect choices were mainly due to object preference. Therefore, the impairment was probably caused by a failure to inhibit inappropriate responses. Together with previous neuropsychological findings, deficits of aged monkeys in the performance of object discriminations can be explained by dysfunction of the frontal cortex.     

Sequential reacquisition as a function of timeout from avoidance

Rats learned to reacquire four similar three-member response sequences. Each sequence member was associated with a different response lever, and the correct sequence of levers (i.e., 3-1-2, 2-1-3, 1-3-2, and 2-3-1) changed each session. The first two correct responses of each sequence postponed shock for a fixed period of time. The third correct response initiated a signalled timeout from avoidance. Incorrect responses did not affect the shock interval or reset the sequence. The effects of manipulating timeout duration on the sequential reacquisition baseline were investigated. All subjects displayed biphasic reacquisition performances similar to those controlled by food. The phases were characterized by an initial increase in accuracy, which reached a stable level during the latter portion of each session. Timeout duration affected rate of sequence completion and shock density, but not percentage of errors. Rate of sequence completion was fastest with intermediate timeouts (15 to 60 sec), and slowest with extreme durations (1 or 120 sec). Shock densities peaked with extreme durations and were at minimum with intermediate timeout values. The percentage of errors was the same across timeout durations. These data extend the generality of sequential reacquisition as a procedure for studying learning, and demonstrate timeout from avoidance to be a controlling variable.     

Pigeons play the percentages: computation of probability in a bird

The ability to compute probability, previously shown in nonverbal infants, apes, and monkeys, was examined in three experiments with pigeons. After responding to individually presented keys in an operant chamber that delivered reinforcement with varying probabilities, pigeons chose between these keys on probe trials. Pigeons strongly preferred a 75% reinforced key over a 25% reinforced key, even when the total number of reinforcers obtained on each key was equated. When both keys delivered 50% reinforcement, pigeons showed indifference between them, even though three times more reinforcers were obtained on one key than on the other. It is suggested that computation of probability may be common to many classes of animals and may be driven by the need to forage successfully for nutritional food items, mates, and areas with a low density of predators.     

Response stereotypy without automaticity in pigeons

Previous research has shown that when pigeons are required to peck each of two keys four times in any order for reinforcement, stereotyped response sequences develop though they are not demanded by the task. The present experiment explored whether the functional value of sequence stereotypy was that sequences became automatized, freeing cognitive resources for other activities. Pigeons were exposed to variants of the task requiring four pecks on each key that differed in the degree of task stringency. Concurrent with the sequence task, pigeons were required to process the color of the keys in order to complete successfully either a matching or a conditional discrimination task. The delay between sequence execution and matching or discrimination choice was varied between 0 and 10 sec. Matching and discrimination accuracy were decreasing functions of delay, but the stringency of the concurrent sequence task had no effect on choice accuracy, suggesting that stereotyped sequences were automatized. However, the matching and discrimination tasks produced substantial disruptions of sequence performance, suggesting that stereotyped sequences were not automatized. The data were taken to suggest that response stereotypy need not imply automaticity.     

The role of error‐correction procedures in the reinforcement of errors

In Study 1 three different error‐correction procedures and a trial‐and‐error procedure were implemented in a Japanese word/phrase receptive identification task. Training procedures differed in the specific type of error that evoked error‐correction feedback. Outcomes indicated that when the error‐correction consequence was contingent on incorrect responses, incorrect responses predominated initially. When this feedback consequence followed no response, non‐responding initially prevailed. Task mastery was achieved in fewer trials with error‐correction procedures than with the no prompt/trial‐and‐error. However, post‐mastery probes 1 week after mastery indicated retention was greater for participants in the trial‐and‐error training condition. These patterns indicated that the feedback prompt itself could be functioning as a positive reinforcer. Demonstration of this effect would have been more definitive had individual participants been exposed to each experimental condition, and the limited number of participants in the trial‐and‐error group weakened conclusions from Study 1. These outcomes necessitated procedural replication. The initial question remained unanswered: if feedback were delivered contingent on trainee request and not contingent on errors, how would this affect errors and retention? Study 2, with a prompt‐request (‘Show me’) answered this question. Copyright © 2006 John Wiley & Sons, Ltd.     

Reinforcement of least-frequent sequences of choices

When a pigeon's choices between two keys are probabilistically reinforced, as in discrete trial probability learning procedures and in concurrent variable-interval schedules, the bird tends to maximize, or to choose the alternative with the higher probability of reinforcement. In concurrent variable-interval schedules, steady-state matching, which is an approximate equality between the relative frequency of a response and the relative frequency of reinforcement of that response, has previously been obtained only as a consequence of maximizing. In the present experiment, maximizing was impossible. A choice of one of two keys was reinforced only if it formed, together with the three preceding choices, the sequence of four successive choices that had occurred least often. This sequence was determined by a Bernoulli-trials process with parameter p. Each of three pigeons matched when p was ½ or ¼. Therefore, steady-state matching by individual birds is not always a consequence of maximizing. Choice probability varied between successive reinforcements, and sequential statistics revealed dependencies which were adequately described by a Bernoulli-trials process with p depending on the time since the preceding reinforcement.     

Response stereotypy without automaticity: Not quite involuntary attention in the pigeon

Previous research has shown that when pigeons are required to peck each of two keys four times in any order for reinforcement, stereotyped response sequences develop, though they are not required by the task. However, despite this stereotypy, sequences are not automatized. They require sufficient attention that sequence performance is disrupted by a concurrent, delayed matching-to-sample task. In the present experiments, pigeons were required to execute response sequences and delayed matching-to-sample concurrently. Experiment 1 showed that the matching task disrupts sequence performance even when the matching sample stimuli do not have to be processed concurrently with sequence execution. The presence of sample stimuli during sequence execution seems to command attention even if the sample stimuli are also available for processing when sequences are not being executed. Experiment 2 used the directed forgetting technique to show that pigeons do seem able to ignore sample stimuli if they are given information that tells them they will not be required to remember those stimuli later. Together, the experiments suggest attentional mechanisms in the pigeon that are not quite involuntary.     

Control of responding by location of auditory stimuli: role of differential and non-differential reinforcement

Sound was presented to monkeys through one of two loudspeakers, each adjacent to a response key. A response on the key adjacent to the sound source was reinforced (correct response). A response on the other key produced a timeout (incorrect response). Under these conditions, over 90% of responses were correct within one or two sessions. When the procedure was changed so that a response on either key was reinforced independently of which speaker was sounding, similar control by location developed within one or two sessions. When conditions were modified by moving the keys away from the immediate vicinity of the speakers, the animals required about 20 sessions to reach a stable level of greater than 90% correct responses under differential reinforcement conditions. No control by location developed under nondifferential reinforcement conditions.     

Effects of GABA modulators on the repeated acquisition of response sequences in squirrel monkeys

The present study investigated the effects of positive and negative GABA(A) modulators under three different baselines of repeated acquisition in squirrel monkeys in which the monkeys acquired a three-response sequence on three keys under a second-order fixed-ratio (FR) schedule of food reinforcement. In two of these baselines, the second-order FR schedule and the discriminative stimuli for the response sequence were manipulated ("chain-strained" and "tandem-strained"). In the third baseline condition, response-independent tail shock was presented during acquisition of the response sequence. All of these baselines maintained high error levels and produced slow rates of acquisition. Under both the chain-strained and tandem-strained conditions, the positive GABA(A) modulator triazolam (0.0032-0.1 mg/kg) and the negative GABA(A) modulators beta-CCE (ethyl-beta-carboline-3-carboxylate; 0.01-1 mg/kg), beta-CCM (methyl-beta-carboline-3-carboxylate; 0.0032-0.1 mg/kg), and FG-7142 (methyl-beta-carboline-3-carboxamide; 0.18-10 mg/kg) dose-dependently decreased overall response rate compared to administration of saline (control). Under the same two conditions, triazolam and the negative GABA(A) modulators also increased the percentage of errors; however, the effects on accuracy frequently depended on the baseline condition and the particular modulator. In contrast, triazolam only decreased errors and enhanced acquisition in the presence of concurrent response-independent tail shock when compared to saline administration under this condition. The neutral GABA(A) modulator, flumazenil (1 mg/kg), had no effect on rate or accuracy of responding when administered alone, but antagonized the rate-decreasing and error-increasing effects produced by the negative GABA(A) modulators. Together, these data suggest that the effects of both the positive and negative GABAA modulators on acquisition can be similar in squirrel monkeys (i.e., both types of modulator may produce rate-decreasing and error-increasing effects) and that their effects on acquisition depend, in part, on the environmental conditions maintaining acquisition.