Incentive contrast in honey bees.

Bees trained to come to the laboratory for a 20% sucrose solution accept it readily, but bees trained with a 40% sucrose solution and tested with the 20% solution show a pattern of interrupted feeding that may last for several minutes. Bees trained with 20% and tested with 40% sucrose are undisturbed. When the animals are offered two samples of the 20% solution simultaneously, they drink to repletion from whichever they first taste on each visit, but if both a 20% and a 40% drop are offered the 20% solution is rejected after a single experience of the 40% solution. Although these results are analogous in many respects to incentive contrast effects found in mammals, they can be understood in sensory terms and do not require the assumption of learning about reward.

Reward and learning in the goldfish.

An experiment with goldfish showed the effects of change in amount of reward that are predicted from reinforcement theory. The performance of animals shifted from small to large reward improved gradually to the level of unshifted large-reward controls, while the performance of animals shifted from large to small reward remained at the large-reward level. The difference between these results and those obtained in analogous experiments with the rat suggests that reward functions differently in the instrumental learning of the two animals.

Puromycin and retention in the goldfish.

A first experiment compared the behavior of goldfish injected with puromycin immediately after each of a weekly series of brief discriminative training sessions in the shuttlebox to that of appropriate controls. Discrimination was not prevented, nor was escape from shock impaired, but probability of response to the conditioned stimuli, both positive and negative, was reduced substantially. These results suggest that puromycin interferes with the consolidation of conditioned fear. The null outcome of a second experiment, in which all training was given in a single long session instead of a series of short sessions, suggests (contrary to recent indications) that consolidation begins in the training session. The conditioned-fear hypothesis is supported by the results of a third experiment in which the animals were shocked upon entering a goalbox to which they had previously learned to swim for food; animals injected with puromycin, immediately after the shock, entered the goalbox more readily 1 week later than did appropriate controls.

Modification of attention in honey bees.

Honey bees were trained in two consecutive two-dimensional (color-position) problems with one dimension (color or position) relevant and the other irrelevant in each problem. As in analogous experiments on dimensional transfer in rats and monkeys, performance in the second problem was more accurate when the relevant and irrelevant dimensions were the same as in the first problem than when they were interchanged. The results of further experiments suggest that the transfer is mediated by different modes of responding that develop in color and position problems rather than by some special process of dimensional selection, such as has been assumed to operate in vertebrates.

Learning in fish with transplanted brain tissue.

Material taken from fish embryos during gastrulation was implanted at prospective tectal sites in host embryos of the same age and species. When mature, the hosts were trained in a series of habit reversals. Two of six animals showed progressive improvement in reversal (a phenomenon not typically found in fish, but characteristic of higher animals), two showed unusually few errors, and two behaved normally. Differences in performance were correlated with differences brain structure.

Interference and forgetting in bird and fish.

Retention in the pigeon and in the goldfish was measured 1 day or 2 weeks after the mastery of each of a series of color discrimination. The amount of forgetting in the pigeon increased with the number of prior problems and increased more rapidly at the longer than at the shorter interval. the amount of forgetting in the goldfish was independent, at both intervals, of the number of prior problems. These results point to the operation of different memory mechanisms in the two animals.

Reversal learning and forgetting in bird and fish.

Pigeons and goldfish were trained in red-green discrimination in daily sessions, with the rewarded color changed every 2 days. Improvement in the performance of the pigeons could be traced to decrements in retention from each day to the next. The goldfish showed no improvement and no decrements in retention. The results suggest that progressive improvement in habit reversal is a product of proactive interference, and that the absence of improvement in the fish is due, not to the lack of some higher-order process which operates to produce improvement in higher vertebrates, but to a difference in learning-retention mechanisms.

Within-subjects experiments on blocking and facilitation in honeybees (Apis mellifera).

It has long been suspected in the vertebrate literature, but demonstrated only recently in work with honeybees (Apis mellifera), that the different treatments of nontarget stimuli in conventional between-groups blocking experiments may give the appearance of blocking independently of experience with the target stimulus. The same difficulty does not arise in within-subjects experiments, and in a series of such experiments with odors and colors free-flying honeybees gave no evidence of blocking; separate reinforced presentations of one element of a reinforced compound failed to reduce responding to the second. There was, however, clear evidence of facilitation; separate nonreinforced presentations of one element of a reinforced compound increased responding to the second. The implications of the results for further work on compound conditioning in honeybees and other animals are considered.

Delayed symbolic matching in honeybees (Apis mellifera).

In a recent experiment on short-term memory (P. A. Couvillon, T. P. Ferreira, & M. E. Bitterman, 2003), honeybees (Apis mellifera) learned to choose between 2 colors on the basis of immediately preceding experience with 1 of them. Some learned to choose the same color as the sample (perseveration or matching), others to choose the alternative color (alternation or nonmatching). Performance in the 2 problems was very much the same. In the present experiment, honeybees learned no less readily to choose between the 2 colors on the basis of sample stimuli that were different from the colors (symbolic matching). A simple associative interpretation of the results is proposed.

The role of contingency in classical conditioning.

The assumption that classical conditioning depends on a contingent relation between the conditioned stimulus (CS) and the unconditioned stimulus (US), which was proposed some decades ago as an alternative to the traditional contiguity assumption, still is widely accepted as an empirical generalization, if no longer as a theoretical postulate. The first support for the contingency assumption was provided by experiments in which occasional CS-US pairings produced no response to the CS in random training--i.e., training in which the probability of the US was the same in the presence and absence of the CS. Those early experiments, the results of which too often are taken at face value, are reconsidered along with various later experiments that show conditioning, both of the CS and its context, in random training. The evidence suggests that CS-US contingency is neither necessary nor sufficient for conditioning and that the concept has long outlived any usefulness it may once have had in the analysis of conditioning.

Blocking and overshadowing in two species of fish.

In Experiment 1, a mixed blocking-overshadowing effect of color on an auditory discrimination was demonstrated in goldfish. In Experiment 2 (with lines differing in color and angle), blocking of angle by color and of color by angle was demonstrated in goldfish, In Experiment 3 (again with lines differing in color and angle), overshadowing of angle by color was demonstrated in carp, but the same animals (like goldfish in a previous study) failed to show greater intradimensional than extradimensional transfer. These results are consistent with the hypothesis that blocking and overshadowing are general phenomena of vertebrate learning. They suggest also that the processes responsible for blocking and overshadowing are different from those which produce the dimensional transfer effect.

Second-order appetitive conditioning in goldfish.

Second-order appetitive conditioning was studied in a series of experiments on goldfish. The first-order conditioned stimuli were white lines of different orientations projected on a target, the second-order conditioned stimuli were diffuse colors projected on the target, the unconditioned stimulus was liquid food delivered through a nipple at the center of the target, and the measured response was target striking. In the first experiment, sustained second-order conditioning was demonstrated. Evidence of colon-line association then was sought without success in two further experiments--a sensory preconditioning experiment in which second-order pairings preceded the first-order pairings, and a first-order revaluation experiment in which responses to second-order stimuli were measured after reversal training with positive and negative first-order stimuli. The stimulus-response reinforcement principle affords the best interpretation of the second-order conditioning demonstrated here: Responding to a second-order stimulus is strengthened by a contiguous first-order stimulus that acquires reinforcing properties on first-order trials.

The two-test strategy in the study of inhibitory conditioning.

It is commonly believed that both a summation test and a retardation test should be used to determine whether a stimulus becomes inhibitory in consequence of some specified treatment, because the 2 tests together rule out alternative interpretations. Depending, however, on the choice of control treatments, a single test may provide credible evidence of inhibition or both together may not. A comprehensive review of the 2-test literature shows that suitable controls have been used only rarely and that compelling evidence of inhibition is correspondingly rare. The only such evidence now available is provided by retardation tests in experiments with some variation of A+/AB- training as the putatively inhibitory treatment.

Quantitative two-process analysis of avoidance conditioning in goldfish.

The shuttlebox performance of goldfish was studied under standardized conditions in a variety of problems--with or without an avoidance contingency, a conditioned stimulus (CS)-termination contingency, and an escape contingency. The effects of CS-only, unconditioned stimulus (US)-only, and explicitly unpaired training were also examined. All the data could be simulated quantitatively with a version of O. H. Mowrer's (1947) 2-process theory expressed in 2 learning equations (1 classical, the other instrumental) and a performance equation. The good fit suggests that the theory is worth developing further with new experiments designed to challenge it.

Exploratory studies of inhibitory conditioning in honeybees (Apis mellifera).

The best available evidence of inhibitory conditioning in vertebrates comes from experiments in which variants of A+/AB- and A+/B- training were compared in terms of response to B in summation and retardation tests, the results suggesting that inhibition is generated by nonreinforcement as an increasing function of the excitatory value of the setting. We report here 7 experiments with foraging honeybees (Apis mellifera) that failed to show a difference in the effects of the 2 treatments. On the basis of previous experiments as well as supplementary experiments whose results give no reason to doubt the sensitivity of the training techniques and measures used, our consistently negative results may mean either that inhibition in honeybees is generated by nonreinforcement independently of the setting or that there is no inhibitory conditioning at all in honeybees--that the only associative function of nonreinforcement is to reduce excitatory strength.

The overlearning-extinction effect and successive negative contrast in honeybees (Apis mellifera).

In three experiments with free-flying honeybees, the previously discovered overlearning-extinction effect was replicated under different conditions and shown to depend on magnitude of reinforcement: The effect appeared in training with a 50% sucrose solution but not with a 20% solution. The results prompted a fourth experiment in which successive negative contrast was demonstrated: The animals were disturbed to find the 20% solution on a distinctive target that always before had been baited with the 50% solution. The conclusion is that the overlearning-extinction effect is an instance of contrast and can be understood in terms of frustration engendered by unrealized anticipation of reinforcement.