Daphna Buchsbaum

The power of possibility: causal learning, counterfactual reasoning, and pretend play

We argue for a theoretical link between the development of an extended period of immaturity in human evolution and the emergence of powerful and wide-ranging causal learning mechanisms, specifically the use of causal models and Bayesian learning. We suggest that exploratory childhood learning, childhood play in particular, and causal cognition are closely connected. We report an empirical study demonstrating one such connection—a link between pretend play and counterfactual causal reasoning. Preschool children given new information about a causal system made very similar inferences both when they considered counterfactuals about the system and when they engaged in pretend play about it. Counterfactual cognition and causally coherent pretence were also significantly correlated even when age, general cognitive development and executive function were controlled for. These findings link a distinctive human form of childhood play and an equally distinctive human form of causal inference. We speculate that, during human evolution, computations that were initially reserved for solving particularly important ecological problems came to be used much more widely and extensively during the long period of protected immaturity.

Do children copy an expert or a majority? Examining selective learning in instrumental and normative contexts

This study examined whether instrumental and normative learning contexts differentially influence 4- to 7-year-old children’s social learning strategies; specifically, their dispositions to copy an expert versus a majority consensus. Experiment 1 (N = 44) established that children copied a relatively competent “expert” individual over an incompetent individual in both kinds of learning context. In experiment 2 (N = 80) we then tested whether children would copy a competent individual versus a majority, in each of the two different learning contexts. Results showed that individual children differed in strategy, preferring with significant consistency across two different test trials to copy either the competent individual or the majority. This study is the first to show that children prefer to copy more competent individuals when shown competing methods of achieving an instrumental goal (Experiment 1) and provides new evidence that children, at least in our “individualist” culture, may consistently express either a competency or majority bias in learning both instrumental and normative information (Experiment 2). This effect was similar in the instrumental and normative learning contexts we applied.

Inferring action structure and causal relationships in continuous sequences of human action.

In the real world, causal variables do not come pre-identified or occur in isolation, but instead are embedded within a continuous temporal stream of events. A challenge faced by both human learners and machine learning algorithms is identifying subsequences that correspond to the appropriate variables for causal inference. A specific instance of this problem is action segmentation: dividing a sequence of observed behavior into meaningful actions, and determining which of those actions lead to effects in the world. Here we present a Bayesian analysis of how statistical and causal cues to segmentation should optimally be combined, as well as four experiments investigating human action segmentation and causal inference. We find that both people and our model are sensitive to statistical regularities and causal structure in continuous action, and are able to combine these sources of information in order to correctly infer both causal relationships and segmentation boundaries.

Children's causal inferences from conflicting testimony and observations.

Preschoolers use both direct observation of statistical data and informant testimony to learn causal relationships. Can children integrate information from these sources, especially when source reliability is uncertain? We investigate how children handle a conflict between what they hear and what they see. In Experiment 1, 4-year-olds were introduced to a machine and 2 blocks by a knowledgeable informant who claimed to know which block was better at activating the machine, or by a naïve informant who guessed. Children then observed probabilistic evidence contradicting the informant and were asked to identify the block that worked better. Next, the informant claimed to know which of 2 novel blocks was a better activator, and children chose 1 block to try themselves. After observing conflicting data, children were more likely to say the informants block was better when the informant was knowledgeable than when she was naïve. Children also used the statistical data to evaluate the informants reliability and were less likely to try the novel block she endorsed than children in a baseline group who did not observe data. In Experiment 2, children saw conflicting deterministic data; the majority chose the block that consistently activated the machine as better than the endorsed block. Childrens causal inferences varied with the confidence of the informant and strength of the statistical data, and informed their future trust in the informant. Children consider the strength of both social and physical causal cues even when they disagree and integrate information from these sources in a rational way.

The effect of experience and rearing environment on the behaviour of crab spiderlings during their first weeks of life

To examine (1) the effect of experience and rearing environment on learning and behaviour over time, and (2) when (or if) newly-emerged young begin to incorporate experience into their choices, we compared the activity levels and hunting-site preferences of newly-emerged laboratory and field-reared crab spiders (Misumena vatia (Clerck, 1757)). We split spiderlings from eight broods into cohorts (1) released on goldenrod (Solidago canadensis) in the field and (2) retained individually in the laboratory, but provisioned with Drosophila melanogaster. We then tested both field-retrieved and laboratory-reared individuals at days 3, 7, 14 and 21 for their (1) rate of activity in cages over 15 min and (2) choice of either goldenrod or wild carrot (Daucus carota), two frequent hunting sites in the field. The rearing environment clearly influenced the spiderlings’ activity levels: field-reared spiderlings became more active than laboratory-reared spiders over time; however, their choice of flowers did not change, even though their experience differed. Thus, innate mechanisms dominated the spiders’ early hunting-site choices, and experience only informed their decisions at later stages in ontogeny. The spiders may, therefore, have multiple mechanisms for learning that begin to operate at different times, since spiderlings improve their locomotor performance through experience, while failing to use early experience in choosing hunting sites at this early point in ontogeny.