Categorical color coding by pigeons

Pigeons were trained on two independent tasks. One involved red and yellow hues, the other involved blue and green hues. For half of the birds, the two tasks were the same (i.e., both tasks were either matching-to-sample, or oddity-from-sample). For the remaining birds, the two tasks were different (i.e., one task was matching-to-sample; the other task was oddity-from-sample). Following acquisition, the pigeons were exposed to test trials on which either the correct or the incorrect comparison hue was replaced with one of the hues from the other task. On yellow-sample trials and on green-sample trials, the pigeons performed as if they had a common code for yellow and green. When there was one comparison available that was appropriate to the “yellow/green” code, performance remained high; but when either both comparisons or neither comparison was appropriate to the “yellow/green” code, performance dropped. The pigeons also tended to code red samples as green and to code blue samples as yellow. The results indicate that pigeons can categorically code colors under conditions that rule out a failure to discriminate among the colors.     

Incentive learning and the motivational control of instrumental performance by thirst

The role of incentive learning in instrumental performance following a shift in the degree of water deprivation was analyzed in three experiments. In Experiments 1A and IB, rats trained to perform an instrumental action reinforced with either sucrose or maltodextrin solutions when in a high-deprivation state were subsequently shifted to a low-deprivation state and tested in extinction. This within-state shift in water deprivation reduced instrumental performance only when the animals had been exposed to the reinforcer in the low-deprivation state prior to instrumental training. In Experiment 2, a concurrent training procedure was used to assess whether the change in the value of the reinforcer brought about by preexposurewas mediated by the contingency between the instrumental action and the reinforcer. Preexposure to the reinforcer under the low-deprivation state produced a selective reduction of the performance of the action upon which it was contingent during training when testing was conducted in extinction following a shift from the high- to the low-deprivation state. These experiments provide evidence that animals have to learn about the incentive value of a reinforcer in a particular motivational state through exposure to the reinforcer in that state.     

Auditorysame/different concept learning by monkeys

Two rhesus monkeys learned the auditory abstract concept ofsame/different. They were trained with 38 different environmental and natural sounds, which were arranged in different combinations as training progressed. Upon transfer to 138 different novel stimuli, they performed as well (78.8% correct) on the first exposure to the novel stimuli as they did (77.3%) with their training stimuli. The comparatively large set of training sounds, contact with the sound source, and a special fading procedure are thought to have contributed to the monkeys’ being able to learn this concept. Implications for species’ similarities/differences in cognitive processing are discussed.     

Teaching the design and interpretation of graphs through computer-aided graphical data analysis

Graphs are one of the primary means of exploration and communication in the practice of science, but students in science laboratories are customarily taught only the low-level mechanics of constructing a single kind of graph when given a table of information. The use of a microcomputer can relieve the drudgery of plotting, allowing students to pursue higher-level issues in the design and interpretation of graphs through repeated “thought experiments.” We introduced computer-assisted graphical data analysis to inner-city high school students with weak math and science backgrounds, emphasizing the dynamic manipulation of various kinds of graphs to answer specific questions. Drawing on extensive recordings and classroom observations, we describe examples of the performance of these students on open-ended problem-solving tasks in which graphs can be used to arrive at meaningful answers to applied data analysis problems.     

Flavor-evaluative conditioning is unaffected by contingency knowledge during training with color-flavor compounds

In two experiments, participants inspected and drank a series of drinks, half of which contained sugar and half unpalatable Tween20 (tween). Each sugar and tween drink had a particular flavor and color. Following this training, the flavors of the sugar drinks were assigned higher hedonic evaluations than were those of the tween drinks, even though the participants did not reliably report which flavors had been present in the sugar and tween drinks during training. Moreover, the evaluative conditioning of the flavors was unaffected by whether or not the colors alone had been pretrained to predict the presence of sugar or tween in the drinks. In accord with Baeyens, Eelen, van den Bergh, and Crombez (1990), we conclude that flavor-evaluative conditioning is not mediated by contingency learning.