Evidence of the conjunction fallacy in Rats (Rattus norvegicus) following two-lever choice training

The conjunction fallacy, a well-known cognitive bias, occurs when individuals erroneously judge the conjunction of 2 events as more probable than either event occurring independently (Tversky & Kahneman, 1983). Although this fallacy has been extensively studied in humans, recent research from our lab shows that nonhuman animals, such as rats, also demonstrate this bias (González, Sadeghi, et al., 2023). In González et al.'s study, rats were trained in a feature-positive/feature-negative go/no-go discrimination (A-/AX+; B+/BY-, where A and B were auditory cues and X and Y were visual cues, and + and-indicate that lever pressing was reinforced or not). At test, cues A and B were presented on probe trials with cues X and Y either unoccluded and off or occluded by a metal shield. Responding was low to A and high to B when the light was visible (and off). When the light was occluded, however, this pattern reversed, with high responding to A and low responding to B. This indicated a bias toward expecting the occluded light to be present rather than absent, consistent with a conjunction fallacy. In a go/no-go procedure, however, some trial types are rewarded, whereas others are unrewarded. This reward asymmetry may bias attention to the different cue conditions. Thus, the current study replicated this conjunction bias using a 2-alternative choice procedure where reinforcement was available on all trial types, removing this potential bias. Our demonstration of a strong conjunction fallacy replicated our prior results, setting the stage for parallel studies in rats and humans. (PsycInfo Database Record (c) 2026 APA, all rights reserved).