|
| Divided Attention and Divided Stimulus Control |
| Tuesday, May 26, 2009 |
| 9:00 AM–10:20 AM |
| North 226 AB |
| Area: EAB/TPC; Domain: Experimental Analysis |
| Chair: Michael C. Davison (University of Auckland) |
| Abstract: This symposium comprises four data-based presentations that investigate compound stimulus control, and the various variables that move stimulus control, or attention, between different dimensions and elements of such compounds. Elliffe uses probe transfer tests to elucidate control by redundant relevant cues in learning a matching-to-sample procedure. Zentall investigates the element superiority effect in compound stimulus control and finds it unaffected by delay to choice, suggesting the effect is an attentional input effect. Shahan and Quick varied the probability of reinforcement for correctly reporting lines and colors in compound stimuli under delay conditions. Even when stimulus discriminability was held equal by varying sample durations, they found the element-superiority effect, again supporting an attentional process. Davison replicates the control of sample responding by relative reinforcer rates for two stimulus dimensions reported previously by Shahan and Podlesnik, and presents a quantitative model for the effect. |
| |
| Selective stimulus control in matching-to-sample |
| DOUGLAS ELLIFFE (University of Auckland) |
| Abstract: Six pigeons learned a redundant-relevant-cues identity matching-to-sample task in which the stimuli differed on two dimensions, color and shape. Probe transfer tests in extinction tried to identify whether either element overshadowed the other by arranging choice stimuli that shared one element and differed on the other. The shared element either matched or differed from that element of the sample on different trials. Color generally overshadowed shape, particularly early in testing and on matching common-element trials. However, continued testing produced both increased control by the weaker dimension and a strong tendency to choose the stimulus element that did not match the sample when the common element of the choice stimuli differed from the sample. This suggests that continued probe testing in extinction resulted in a loss of conditional control by the sample, leading to consistent choice of the “wrong” comparison stimulus. Supporting this interpretation, a series of resistance-to-reinforcement transfer tests reversed the trend and produced clear overshadowing of shape by color. |
| |
| The element superiority effect: An attentional phenomenon that occurs at input rather than at output |
| THOMAS ZENTALL (University of Kentucky) |
| Abstract: When pigeons are presented with a compound sample (color and line orientation) and tested with comparisons from either dimension, matching accuracy is poorer than when the sample is either color or line alone (the element superiority effect). We tested the hypothesis that this difference resulted from impulsive choice at test by asking whether the effect would be reduced when (1) choice latency was naturally longer, (2) a delay was required prior to comparison choice (a brief FI or DRL replaced CRF as comparison choice), (3) a more “natural” delay was imposed by presenting the sample at the back wall of the chamber and the comparison choice at the front wall. In all cases a constant element superiority effect was found. Thus, the effect appears to be attentional in nature resulting from a deficit at input rather than at output. |
| |
| Reinforcement probability affects adjusting sample duration in a divided-attention task. |
| TIMOTHY A. SHAHAN (University of Utah), Stacey Quick (Utah State University) |
| Abstract: Pigeons responded on a delayed matching-to-sample procedure with intermixed compound (color + line orientation samples) and single-element (color or line samples) trials. Both trial types were followed by element comparisons comprised of either two colors or two lines for which accurate matches resulted in reinforcement with the same probability (i.e., color:line reinforcement ratio always =1). Across conditions, the overall probability of reinforcement was varied while accuracy on compound and element trials was held constant and equal by adjusting sample durations within sessions. Replicating previous work, required sample durations were longer on divided attention trials than on single-element trials (i.e., the element superiority effect). In addition, obtained sample durations for both compound and single-element trials increased with decreases in reinforcement probability. This result is consistent with the suggestion that the effects of reinforcement on conditional discrimination accuracy are mediated by changes in attending to the sample stimuli. However, the impact of reinforcement probability on required sample durations does not appear to depend upon whether attention is divided or not. |
| |
| A model for divided stimulus control |
| MICHAEL C. DAVISON (University of Auckland) |
| Abstract: Four pigeons were trained on a compound conditional discrimination using two flash frequencies and two stimulus durations. Over a series of 7 conditions, the probability that correct detections of stimuli on each of the two dimensions would be reinforced was varied from 0 to 1.0. Stimulus discriminability measured by log d was a direct function of the probability of reinforcement, replicating previous research by Shahan and Podlesnik. We developed a quantitative model, based on the discriminative law of effect (Davison & Nevin, 1999) that accounted nicely for the way in which reinforcers produce divided stimulus control. |
|
| |