Leonie Baumann, Lydia Paulin Schidelko, Marina Proft, Tanya Behne and Hannes Rakoczy
Affiliation: University of Göttingen
Category: Psychology
Keywords: Modal cognition, Agentive modality, Possibility reasoning, Uncertainty
Date: Tuesday 2nd of September
Time: 18:00
Location: Maria Skłodowska-Curie Hall (123)
View the full session: Agency
Modal reasoning – thinking about what can, must, or may be the case – is an essential form of human cognition. It is foundational for virtually all kinds of higher thought, such as causal, moral, counterfactual, or probabilistic thinking. Recently, there have been big debates in comparative and developmental psychology whether such modal reasoning capacities may be uniquely human, potentially linked to language, and when and how in ontogeny they may emerge (Engelmann et al., 2021; Leahy & Carey, 2020; Redshaw & Ganea, 2022; Redshaw & Suddendorf, 2020). Empirically, the situation is complex and confusing: on the one hand, numerous studies indicate that both young children and great apes have major difficulty reasoning about multiple incompatible possibilities in direct tasks (e.g., Beck et al., 2006; Engelmann et al., 2023; Leahy, 2023; Mody & Carey, 2016; Redshaw & Suddendorf, 2016; Robinson et al., 2006). These findings have been interpreted as evidence for a principled competence deficit: non-human primates and young children lack the conceptual capacity for modal thought proper (Carey et al., 2020). These conservative accounts assume that modal reasoning emerges relatively late, not before age 4, possibly based on the acquisition of modal language or meta-representational thought (Leahy & Carey, 2020; Redshaw & Suddendorf, 2020). However, a different set of findings suggests that the failure of young children and apes in modal reasoning tasks may not reflect a competence deficit; rather, these failures may constitute false negatives that are explainable by performance limitations (Phillips & Kratzer, 2024). Once tasks are suitably modified, much younger children may already perform proficiently. For example, 3-year-olds seemed to reason effortlessly about multiple incompatible possibilities in a recent study (Alderete & Xu, 2023). In light of these mixed findings, Phillips and Kratzer (2024) argue that the conceptual core competence for representing incompatible possibilities may develop quite early in human ontogeny and may be present in other species. It may, however, be masked in some modal reasoning tasks due to additional cognitive performance demands (to do with meta-cognition, naïve physics, etc.). Against the background of the recent debates and mixed empirical findings, the present studies address the question when true modal reasoning competence emerges. We aim to contribute to clarifying why children pass some modal reasoning tasks and fail others. To this end, we test whether modal reasoning proficiency can be uncovered earlier with suitably simplified and modified methods. Conceptually, we build on recent theoretical work that suggests that agentive modality is a crucial foundation for modality in general and should thus matter developmentally (Maier, 2015; Vetter, 2023). Agentive modality – thinking about what one can, may or is able to do – is related to our abilities and opportunities as agents (List, 2023; Maier, 2015; Vetter, 2023). Applied to present purposes, children may generally find it easier to reason about what they might do rather than what might be the case. Tasks where children act themselves may pose less demands than tasks where children react to the actions of an experimenter or make detached judgments about possible external events (Phillips & Kratzer, 2024). The present research question is thus whether a focus on agentive modality makes modal reasoning tasks easier and uncovers modal reasoning competences earlier in ontogeny. To address this, we enriched modal reasoning tasks with an agentive structure. In Study 1, sixty 3- to 4-year-old children received two tasks where they chose between performing an action that must lead to a reward and an action that might lead to a reward. The agentive coins task required children to toss a coin with a cup. In the test trials, the target coin had the winning color on both sides, whereas the alternative coin was a fair coin with one winning and one losing side. The agentive slides required children to drop a marble into one of two slides. In the test trials, the target slide resulted in winning in any case, whereas the alternative forked slide sometimes led to winning and sometimes to losing. In both tasks, children had to represent that performing the target action must result in winning, whereas performing the other action might result in winning or losing. Results show that 3- and 4-year-old children perform competently in the agentive coins task but not in the agentive slides task. This raises the question why some tasks with an agentive structure are solvable for younger children whereas other tasks remain difficult despite agentive modifications. In Study 2, we set out to explore the effect of an agentive task structure more directly by comparing an agentive and a non-agentive condition within the same modal reasoning task. More generally, Study 2 also systematically explored other factors that may play a role in the general pattern of successes and failures of modal reasoning tasks, such as the type of uncertainty. Sixty-two 3- to 4-year-old children received two tasks. In a non-agentive version of the coins task, we manipulated the uncertainty under which children had to reason about incompatible possibilities (epistemic vs. physical uncertainty). In the test trials, children chose between one of two coins either before the coins were tossed by the experimenter or after they had been tossed but the result remained unknown. In the slides task, we manipulated the agentive structure of the task. In the test trials, children either performed the action of dropping the marble themselves or instructed the experimenter into which slide she should drop the marble. Results show that even 3-year-old children perform competently in certain tasks where they have control over the realization of the outcome, irrespective of whether they perform the action themselves and irrespective of the type of uncertainty. The agentive task structure, at least in the way it was operationalized in our tasks, does not seem to have a clear effect on children’s modal reasoning performance. This raises the question whether forced-choice paradigms can truly target the notion of agentive modality. All in all, our findings suggest that modal reasoning competence possibly emerges earlier than assumed by conservative accounts.