RETENTION OF ORDINAL POSITION INFORMATION WITH LIMITED AND EXTENDED SERIAL TRAINING.
Just as in the development of theoretical models for human serial learning, realization of the importance for animals of information about the ordinal position of items in a list followed an initial theoretical emphasis on the role of the memories of list items. Though position learning is now emphasized (e.g., Henson, 1998) in human cognition, it began receiving attention (e.g., Bower, 1971; Ebenholtz, 1972) long after memory-association theories (e. g., Ebbinghaus, 1913) had developed rich traditions. The study of memories, usually of differing reward events, in animal serial learning also has a long history that has produced clear evidence for animals employing reward memories in serial problems; and the understanding of the nature of those memories is now fairly sophisticated (reviewed by Capaldi, 1994). Evidence that animals use information about the position of items in a list is recent, but also clear (Burns, Dunkman, & Detloff, 1999; Burns, Kinney, & Griddle, (2000); Burns, Wiley, & Payne, 1986; Chen , Swartz, & Terrace, 1997; Couvillon, Brandon, Woodard, & Bitterman, 1980; D'Amato & Colombo, 1988, Roiblat, Pologue, & Scopatz, 1983; Straub & Terrace, 1981). Our understanding of the nature of position information for animals, however, is not sophisticated.
In a recent experiment with rats (Burns et al., 2000; Experiment 2) three-trial series were examined in which the reward type differed on each of the three trials. Sucrose (S) or plain (P) Noyes pellets were given on the first and last trials, and no reward (N) was administered on the middle trial. The series were arranged so that both reward memory and ordinal position were valid predictors of the reward outcome on a given trial. For example, an animal may have received the series SNS' and PNS', or PNP' and SNP' (where S' and P' are twice the magnitude of S and P). In these series, the memory of N could be strongly associated with an upcoming reward, and the memory of reward could be associated with N (Capaldi, 1994). Ordinal position could also be associated with reward outcomes because the first and third positions are always rewarded and the second position is never rewarded (Burns et al., 1999). After 38 days of training, the animals were shifted to test series in which ordinal position cues were not alt ered, but reward memory cues were (NNN, SSS, and PPP). In this procedure, the well-established pattern of running fast on the first and third trials and slow on the second trial was retained in all three kinds of series in the transfer tests, a result that shows the influence of learning about position information, but tells us little about the nature of that information.
Burns et al. (2000) offered several possible sources of information about position that included: intentional or unintentional trial-to-trial variations in running procedures (Burns et al., 1986), the time (e.g., Church & Broadbent, 1991) from the first trial to each of the remaining trials in the series, the memory of reward outcomes from previous trials (Capaldi, 1994), and the number of previous trials in the series (e.g., Capaldi & Miller, 1988). It may also be that position fashions a linear, cognitive representation (D'Amato, 1991), which the animal may scan from position to position.
Another problem for clarification concerns the way in which position and reward memories control behavior. Capaldi, Alptekin, Miller, and Birmingham (1997) have proposed that reward memories are decidedly salient cues for hungry animals, and will control behavior even when position information is an equally valid predictor. In one of their experiments, a group trained with position, but not reward memory, as a relevant cue for predicting no reward failed to learn to approach slowly on the never-rewarded trial with 18 days of training. Each rat was trained in the runway on the series PSN, SSN, PPN, and SPN. In another experiment, however, animals trained on either SPN or PSN, series in which both reward memory and position are relevant, learned to approach N slowly after only 15 days of training. That training with both reward memory and position relevant was effective, whereas training with only position relevant was ineffective, could be taken as evidence in favor of the proposition that reward memories are salient, relative to position, in serial learning.
Other findings, however, pose problems for the salience interpretation. For example, Burns et al. (2000; Experiment 1) found that training with PSN, SSN, PPN, and SPN produces equally strong discriminative responding, compared to SPN or PSN, if training is continued to 32 days. Furthermore, Burns et al. shifted their animals to NNN on Day 33 and found that the rats from both training conditions continued their pattern of approaching the first two trials of the series rapidly while approaching the final trial slowly, even though the reward memories had been changed. This finding suggests that position, not reward memory, was controlling behavior in the test.
The present experiment was designed to evaluate a possible reason for the differences in the potency of position cues and reward memories reported by Burns et al. (2000) and Capaldi et al. (1997). Although Capaldi et al. were not concerned with this discrepancy, as we are here, we borrowed an idea reported by them and attributed to an anonymous reviewer. The procedures employed to train rats in a runway include inherently variable intervals among the trials of a series because the intervals are affected by the consistency of running by the animals. Early in training, the intervals from the first to second, and second to third, trials in a three-trial series will be variable, due to inconsistent running. Later in training, the running and intervals stabilize. If the temporal interval from the first trial is functioning as a position cue, one of the possible position cues suggested by Burns et al. and Capaldi et al., it would not be a reliable cue early in training, but it would be reliable late in training.
We trained rats in a runway on the series SNP', a series for which both reward memory and ordinal position are perfectly correlated with reward outcomes. On Day 19 of training, when the rats first began approaching the goal slowly on the second, nonrewarded trial, a matched half of the animals were shifted to NNN for 1 test day. As a control, the remaining animals continued their training with SNP', as did the shifted animals after the test. On Day 39, after the pattern of rapid approach on the first and third trials, and slow approach on the second, was firmly established, the remaining animals were shifted to NNN. The earlier-shifted animals continued as controls. If the time from the first trial is an important position cue and is linked to the differences reported by Burns et al. (2000) and Capaldi et al. (1997), this shift after only limited training should produce results expected if reward memories, not position cues, were controlling behavior. Approach should be rapid on all three of the shift trials, If temporal position cues come to prevail over reward memories with extended training, the pattern of behavior expected in the latter shift should parallel that observed both prior to the shift, and for unshifted controls.
The subjects were 10 experimentally naive male rats, approximately 70 days old, obtained from the colony maintained at the Sunny Hill Pet Center, Cape Girardeau, MO.
The apparatus (described fully in Burns et al., 2000) was an L-shaped wooden runway, painted flat black with a vented clear plastic cover. Start and retrace doors were manually operated, and the animal's running was recorded on a silent digital clock when a photobeam, 196.6 cm beyond the start door, was crossed. Rewards were delivered in an aluminum cup mounted in the goal area.
The rats arrived in the laboratory 2 weeks prior to the beginning of exploration training. During this time, subjects were housed in individual living cages and given unlimited access to Purina Rodent Chow and water. The animals were allowed about 1 hr each day in a group interaction box and were handled about 2 min each during this period. After this 2-week period, food was restricted to 18 g per day.
Exploration consisted of 4 days of runway exposure. Each rat spent two 5-min periods in the runway each day. On Days 1 and 3, six .045-g Noyes pellets (P) of the standard formula were scattered throughout the runway during the first of the two explorations. During the second daily exploration, six .045-g Noyes pellets of the sucrose formula (S) were scattered. On Days 2 and 4 of exploration, the order of reward was reversed, first S, then R On Days 3 and 4 of exploration, the start and retrace doors were opened and closed several times, 2.5 min into the exploration, to acquaint the animals with these movements.
During experimental training each rat received two SNP' series each day. The first trial of the series terminated in a sucrose (S) reward of four .045-g sucrose-formula Noyes pellets. The terminal trial was rewarded with eight .045-g standard-formula Noyes pellets (P'), and the middle trial (N) was terminated with a 30-s confinement in an unbaited goal area. A trial began with the opening of the start door about 2 s after placing the rat in the start area, and it ended when the rat had consumed the reward or completed the 30-s confinement. Each rat completed all three trials of the series in succession, a procedure that produces an interval between trials of about 20 s. Because the second daily series was not begun until every rat had completed the first series, the interval between series was about 20 min. The order of running of the animals was a random determination made daily.
By Day 19 of experimental training, statistical analyses, explained in the results section, showed that the animals had begun approaching the goal slower on the middle trial of the series than on the initial and terminal trials. The animals were then matched on the basis of the difference between their running times on Trials 2 and 3 and were assigned to one of two conditions. In one condition, Day 19 consisted of administering the two daily series with the same procedure as on the previous days except all trials ended in nonreward (NNN). The other condition was simply a continuation of training with SNP', as a control. On Days 20 - 38, all the rats were trained on the SNP' series. On Day 39, all the animals not shifted on Day 19 were shifted to NNN, while the remaining animals continued SNP' training.
All running times were subjected to a normalizing transformation [x = log(x)] for analysis. As expected, the course of acquisition was characterized by running times that varied significantly over days for the first 18 days of training, F(17, 153) = 19.46, p [less than] .001. The left panel of Figure 1 shows performance on each of the three trials of the series for all the animals after limited training during Days 14 - 18. The figure also shows the performance of the rats shifted to the NNN series on Day 19 and the matched controls that received the SNP' series on Day 19. Using the Bonferroni t, p [less than] .05, to evaluate a reliable effect of trials, F(2, 18) = 31.77, p [less than] .001, it was evident that during the 5 days preceding the shift, the rats approached the goal significantly more slowly on Trial 2 (M = 1.74, SE = .07) than on Trials 1 and 3 (Ms = 1.11 and 1.09, SEs = .03 and .04). On the transfer day following limited training (Day 19), the pattern of approach was the same as it had been on the previous days, and that pattern did not differ (F [less than] 1) between rats shifted to NNN and those continuing on SNP'. As was the case prior to the shift, rats in both the shifted and control conditions approached the goal significantly more slowly [F(2, 16) = 8.02, p [less than] .001: Bonferroni t, p [less than] .05] on the second series trial (M = 1.79, SE = .19) than on the first and third trials (Ms = 1.10 and 1.06, SEs = .07 and .08).
Extended training with the SNP' series was characterized by stable performance over days, F(18, 162) = 1.66, p[greater than] .05, and an improved pattern of approach in both groups (F[less than] 1). The right panel of Figure 1 shows the average times for all animals during Days 34 - 38 and for both shifted (NNN) and control (SNP') animals on Day 39. On the days preceding the extended-training shift, the animals ran significantly slower [F(2, 16) = 124.96, p [less than] .001; Bonferroni t, p [less than] .05] on the second trial (M = 2.61, SE = .06) than on the others (Ms = .94 and .82, SEs = .02). There was no difference (F [less than] 1) between shifted animals and controls on Day 39, and the rats ran faster [F( 2,16) = 74.49, p [less than] .001; Bonferroni t, p [less than] .051 on Trials 1 and 3 (Ms = .85 and .87, SEs = .03 and .05) than on Trial 2 (M= 2.68, SE= .14).
Examining relative control by reward memories and position cues, using transfer tests in which reward memories were changed without altering position cues, produced clear evidence for control by position cues. The pattern of approaching the goal rapidly on the first and third trials and slowly on the middle trial of the SNP' series was retained in transfer tests to NNN administered after both limited and extended training.
If reward memories controlled behavior in these tests, the animals should have approached the goal rapidly on the first trials of the NNN series because experience with the changed reward, and consequent changed memory of reward, had not yet occurred. The experience of N following the first test trial, however, should have retrieved the memory of P established during experimental training. The memory of P should have caused rapid Trial-2 approach by rats anticipating a large quantity of reward. Trial 3 should have shown the same rapid approach for the same reason. But the transfer pattern of rapid approach on all three trials of the NNN series did not occur.
If the rats associated reward with the first and third ordinal positions and nonreward with the middle position, performance in the three-position, NNN tests should have been the same as in the three-position, SNP' training series. This pattern was the pattern obtained after shifts with both limited and extended training. Because running times did indeed vary in the early training prior to the limited-training transfer test, but were stable during the period preceding the extended-training test, temporal position cues should have been effective only with extended training. The finding that position cues controlled behavior after both limited and extended training suggests the possibility that temporal cues are less important in position learning than previously supposed (e.g., Burns et al., 2000; Capaldi et al., 1997). Although it has been argued (e.g., Broadbent, Church, Meck, & Rakitin, 1993) that number and time discriminations are related processes, counting series trials seems to be a more reasonable so urce of position information here than does timing. Of course, several other possible sources of position information, such as cognitive representation of the positions (D'Amato, 1991), need to be evaluated.
Capaldi et al. (1997) failed to obtain evidence of position learning with series that varied the two initial reward values in three-trial series for which the third position was never rewarded. These researchers, however, provided only limited training. Examining the same series as Capaldi et al., Burns et al. (2000) found clear evidence of position learning with extended training. One possible interpretation of the differing results is that position learning is an extended-training process (Capaldi et al., 1997), but the present finding that behavior patterns were retained in the NNN test following both limited and extended training with the SNP' series is not compatible with this interpretation.
Reward memories may function as cues independent of position, but they may also be one source of position cues, to the extent that the memories are correlated with ordinal position (Burns et al., 1999; Burns et al., 2000). It is possible that the method of varying reward values on Trials 1 and 2 while holding nonreward constant on Trial 3 (Burns et al., 2000; Capaldi et al., 1997) produces a more difficult position discrimination than the method employed in the present research, which holds both reward and position constant in the SNP' series. For this reason, we may have found position learning with both limited and extended training. Another possibility (E. J. Capaldi, personal communication, January 13, 2000) is that rats learn on the basis of both position and reward memory in any situation for which these cues are relevant. According to this view, the transfer test to NNN that we employed favors the expression of position learning, whereas other transfer tests, which alter position information while mai ntaining the integrity of reward memories, show evidence of reward learning.
This research was supported by a Grants and Research Funding grant from Southeast Missouri State university.
Correspondence concerning this work should be addressed to Richard A. Burns, Department of Psychology, Southeast Missouri State University, One university Plaza, Cape Girardeau, Missouri 63701. (E-mail: firstname.lastname@example.org).
BOWER, G. H. (1971). Adaptation-level coding of stimuli and serial position effects. In M. H. Appley (Ed.), Adaptation-level theory. New York: Academic Press.
BROADBENT, H. A., CHURCH, R. M., MECK, W. H., & RAKITIN, B. C. (1993). Quantitative relationships between timing and counting. In S.T. Boysen & E. J. Capaldi (Eds.), The development of numerical competence: Animal and human models. Hillsdale, NJ: Lawrence Erlbaum Associates.
BURNS, R. A., DUNKMAN, J. A., JR., & DETLOFF, S. L. (1999). Ordinal position in the serial learning of rats. Animal Learning & Behavior, 27, 272-279.
BURNS, R. A., KINNEY, B. A., & CRIDDLE, C. R. (2000). Position cues and reward memories as compatible components of serial learning. Learning & Motivation, 31, 236-250.
BURNS, R. A., WILEY, L. R, & PAYNE, T. L. (1986). Temporal cuing of runs in series of reward reduces interevent anticipation. Animal Learning & Behavior, 14, 190-196.
CAPALDI, E. J. (1994). The sequential view: From rapidly fading stimulus traces to the organization of memory and the abstract concept of number. Psychonomic Bulletin & Review, 1, 156-181.
CAPALDI, E. J., ALPTEKIN, S., MILLER, D. J., & BIRMINGHAM, K. M. (1997). Is discriminative responding in reward outcome serial learning mediated by item memories of position cues?. Learning and Motivation, 28, 153-169.
CAPALDI, E. J., & MILLER, D. J. (1988). Counting in rats: Its functional significance and the independent cognitive processes that constitute it. Journal of Experimental Psychology: Animal Behavior Processes, 14, 3-17.
CHEN, S., SWARTZ, K. B., & TERRACE, H. S. (1997). Knowledge of ordinal position of list items in rhesus monkeys. Psychological Science, 8, 80-86.
CHURCH, R. M., & BROADBENT, H. A. (1991). A connectionist model of timing. In M. L. Commons, S. Grossberg, & J. E. R. Staddon (Eds.), Neural network models of conditioning and action (pp. 225-240). Hillsdale, NJ: Lawrence Eribaum Associates.
COUVILLON, P. A., BRANDON, S., WOODARD, W. T., & BITTERMAN, M. E. (1980). Performance of pigeons in patterned sequences of rewarded and nonrewarded trials. Journal of Experimental Psychology: Animal Behavior Processes, 6, 137-154.
D'AMATO, M. R. (1991). Comparative cognition: Processing of serial order and serial pattern. In L. Dachowski & C. F. Flaherty (Eds.), Current topics in animal learning: Brain, emotion, and cognition (pp. 165-185). Hillsdale, NJ: Erlbaum.
D'AMATO, M. R., & COLOMBO, M. (1988). Representation of serial order in monkeys (Cebus apella). Journal of Experimental Psychology: Animal Behavior Processes, 14, 131-139.
EBENHOLTZ, S. M. (1972). Serial learning and dimensional organization. In G. H. Bower (Ed.), The psychology of learning and motivation (Vol. 5). New York: Academic Press.
EBBINGHAUS, H. (1913). Memory. (H. A. Ruger & C. E. Bussenius, Trans.). New York : Teachers College, Columbia University. (Original work published in 1885)
HENSON, R. N. A. (1998). Short-term memory for serial order: The Start-End Model. Cognitive Psychology, 36, 73-137.
ROIBLAT, H. L., POLOGE, B., & SCOPATZ, R. A. (1983). The representation of items in serial position. Animal Learning & Behavior, 11, 489-498.
STRAUB, R. O., & TERRACE, H. S. (1981). Generalization of serial learning in the pigeon. Animal Learning & Behavior, 9, 454-468.
|Printer friendly Cite/link Email Feedback|
|Author:||BURNS, RICHARD A.; CRIDDLE, CATHRYN R.|
|Publication:||The Psychological Record|
|Date:||Jun 22, 2001|
|Previous Article:||MATCHING-TO-SAMPLE AND RESPONDENT-TYPE TRAINING AS METHODS FOR PRODUCING EQUIVALENCE RELATIONS: ISOLATING THE CRITICAL VARIABLE.|
|Next Article:||ACQUIRED EQUIVALENCE OF CUES IN LEARNING A MATCHING-TO-SAMPLE TASK BY RATS.|