Suboptimal choice and initial‐link requirement

Abstract

Pigeons (n = 14) were trained in a concurrent-chains suboptimal choice procedure that tested the effect of an increased ratio requirement in the initial links. Fixed-ratio 1 and 25 conditions were manipulated within subjects in a counterbalanced order. In all conditions, distinct terminal-link stimuli on a suboptimal alternative signaled either primary reinforcement (20% of the time) or extinction (80% of the time). On an optimal alternative, two distinct terminal-link stimuli each signaled a 50% chance of primary reinforcement. Preference for the suboptimal alternative was significantly attenuated, and in some birds completely reversed, by the larger response requirement irrespective of condition order. This larger response requirement also generated a notable increase in between-subject variability. A measure of cumulative choice responding is introduced to mitigate the problems associated with traditional session averages. Ordinal predictions of some current theories of suboptimal choice are also considered in light of the results.

Delays to food-predictive stimuli do not affect suboptimal choice in rats.

“…However, there was a high degree of individual-subject variability in the influence of IL duration on suboptimal choice, with 3 of 5 pigeons still preferring the suboptimal alternative to some extent with longer IL durations. Pisklak et al (2019) assessed choice between a suboptimal alternative delivering food 20% of the time with differential TL stimuli and an optimal alternative delivering food 50% of the time without differential TL stimuli. The IL schedule was either an FR 1 (short duration) or an FR 25 (long duration).…”

“…However, this is based on only a few experiments (i.e., Experiment 3 from Dunn & Spetch, 1990; Kendall, 1974; Zentall et al, 2017), some of which have been criticized for unorthodox procedural details (e.g., the use of dark keys as TL stimuli in Kendall, 1974). Furthermore, there is sometimes a high degree of individual-subject variability in the influence of IL duration on suboptimal choice in pigeons (e.g., Pisklak et al, 2019; Zentall et al, 2017), and longer IL durations do not always decrease suboptimal choice (e.g., Experiment 1 from Dunn & Spetch, 1990). Given such limited and inconsistent evidence, some theories do not explicitly include a role for IL duration in suboptimal choice, such as the predictive-value hypothesis (Zentall, 2016) and the associability-decay model (Daniels & Sanabria, 2018).…”

“…Furthermore, there is some evidence suggesting that long TL durations do not increase suboptimal choice when IL duration is relatively long (Experiment 2 from . However, longer IL durations do not always decrease suboptimal choice in pigeons (e.g., Experiment 1 from , and the influence of IL duration on suboptimal choice can be inconsistent across pigeons (e.g., Pisklak et al, 2019;Zentall et al, 2017). Given the uncertainty surrounding such effects, some theories of suboptimal choice do not include a role for IL duration (e.g., Daniels & Sanabria, 2018;Zentall, 2016).…”

“…While the evidence is clear that suboptimal choice increases with longer TL durations, it is currently less clear whether IL duration influences suboptimal choice. Limited evidence suggests that longer IL durations, via longer IL interval schedules (e.g., Experiment 3 from Dunn & Spetch, 1990; Kendall, 1985; Zentall, Andrews, & Case, 2017) or larger IL ratio schedules (Pisklak, McDevitt, Dunn, & Spetch, 2019), decreases suboptimal choice in pigeons. Furthermore, there is some evidence suggesting that long TL durations do not increase suboptimal choice when IL duration is relatively long (Experiment 2 from Dunn & Spetch, 1990).…”

Suboptimal choice: A review and quantification of the signal for good news (SiGN) model.

“…Many recent studies have tested pigeons’ choice between signaled 20% and unsignaled 50% food as shown in Panel B of Figure 1. The key features of these procedures are FR 1 initial links (i.e., a single choice response) and FT 10-s terminal links (i.e., delays fixed at 10 s; Fortes et al, 2016, 2017, 2018; González et al, 2020b; González-Torres et al, 2020; López-Tolsa & Orduña, 2021; Macías et al, 2021; Pisklak et al, 2019b; Stagner et al, 2012; Stagner & Zentall, 2010; Zentall & Stagner, 2011). Results have consistently shown strong preference for the suboptimal (20%) alternative.…”