Association for Behavior Analysis International

We investigated the effect of prior investment on choice in a concurrent-chains procedure with pigeons. Initial links were VI15–VI15 and terminal links were FR15–FR15. Two types of trials in each session, signalled by red and green, were identical except for one aspect. In red trials, the initial link was preceded by an initial investment of 20 pecks on the left key, whereas in the green trials the initial link was preceded by an initial investment of 20 pecks on the right key. If there is an effect of prior investment on choice in the direction predicted by the sunk cost error and the work-ethic effect, left should be preferred on red trials and right on green trials. Conversely, equal preference would result if there is no effect of prior investment on choice, as shown by studies that unsuccessfully tried to replicate the work-ethic effect. The results of several experiments in which we manipulated size of the initial investment, and the relative size of the terminal-link FR have a bearing on both the sunk cost error and the work-ethic effect.

When given a choice between varying and repeating response sequences, humans and nonhumans tend to choose the repeat alternative, mainly when the vary one is highly demanding. Considering that rigorous vary requirements generate sequences with several switches between operanda, the present study investigated the influence of the number of switches (response cost) upon choice. College students were required to choose between two alternatives. In Experiment 1, sequences had to differ from the previous five ones (VAR contingency) and include two (VAR 2 alternative) or five (VAR 5 alternative) switches. In Experiment 2, reinforcers were contingent to the emission of a unique sequence (REP contingency) with two (REP 2 alternative) or five (REP 5 alternative) switches. In Experiment 3, choice was between one vary (VAR 2 or VAR 5) and one repeat (REP 2 or REP 5) contingency. Participants preferred two to five switches under both the VAR and REP contingencies of experiments 1 and 2. Preference for the REP alternative in Experiment 3 increased as a direct function of the number of switches in the VAR alternative. These results suggest that response cost may ascribe aversive properties to vary contingencies, thus affecting choice between varying and repeating response sequences.

Research has shown that choice adapts rapidly to dynamic changes in the relative rate of food delivery arranged by concurrent random interval schedules. Most theories of choice, however, would predict exclusive preference for one of two alternatives providing food according to concurrent random ratio schedules. We explored this possibility in a choice situation that arranged 7 pairs of probabilities in two levers (0.16-0.02, 0.16-0.04, 0.16-0.08, 0.16-0.16, 0.08-0.16, 0.04-0.16, & 0.02-0.16), defining 7 unsignaled components that occurred in random order within the session. Each component arranged ten-food deliveries and was followed by a 1-minute black out. Choice adapted rapidly to dynamic changes in the relative probability of food delivery that occurred in the levers. Although the rats showed a bias for the left lever, pressing on the right lever tracked the probability of food delivery in that lever. The generalized matching law provided of good fits to the logs of response ratio explaining more than 80 percent of the variability. An almost perfect match between obtained and arranged food ratio was observed. Overall preference favored with more responses the lever with the highest probability of food delivery. The implications of these results to theories of choice will be discussed.

The effect of various sounds on the behaviour of domestic hens was examined. The effect of white noise on hens' performance under multiple-concurrent schedules was compared with the effects of an alarm call, the sounds of hens feeding, and a food call. All sound biased hens' responding away from keys associated with them, and the magnitude of this bias was largest for white noise and the food call. White noise suppressed the responding of hens when it was present, however, the other sounds did not. Next, a concurrent-chains procedure was used to assess the same hens' preferences for the same sounds. In the equal initial-links (with no sounds played) hens chose between "sound" and "no sound" terminal-links. Initial-link response biases were away from the key associated with sound for the food call for all hens and the directions of the other biases varied over hens. Terminal-link entry pauses tended to be longer in terminal-links associated with sounds where the longest pauses were with white noise and the food call. Thus, the two procedures gave some similar and some conflicting results. The various measures of behaviour, including response biases, response rates, and terminal-link pauses, will be compared and discussed.