Fig. 1

State-dependent valuation learning. a Preference and state. Starlings previously trained to peck for food at one stimulus (H) while hungry and a different one (PF) when pre-fed were given choices between the stimuli, while being in either state. PF’s delay to food was always 10 s, but H’s was 10, 12.5, 15 or 17.5 s, according to treatment. Data plotted separately according to testing state (white and grey bars) or pooled (black bars). The birds preferred stimuli previously experienced while hungrier, but were sensitive to how much loss (extra delay) this preference caused. Adapted with permission from Pompilio and Kacelnik, Animal Behaviour (2005)—Fig. 2—https://doi.org/10.1016/j.anbehav.2004.12.009. b Preference and cost. Starlings were trained separately with options that differed in the event preceding the outcome-triggering response. Differences were in either the delay (Exps. 1 and 2) or the amount of work (Exp. 3—at criterion and after extensive training). After training, they were given choices between signals for the outcome-triggering responses without having to wait or work for it. In Experiments 1 and 2, there were two options differing in delay, while in Experiment 3, there were three options differing in a required number of travel flights, leading to three different pairwise choices. The birds were indifferent when treated with delays, but preferred outcomes that were typically costlier when treated with effort. Bars represent the between-subjects average (± SEM) of individual proportions of choices. Asterisks indicate a significant (P < 0.05) difference from random. Adapted with permission from Aw, Vasconcelos and Kacelnik, Animal Behaviour (2011)—Fig. 4—https://doi.org/10.1016/j.anbehav.2011.02.015