Search

Eating onion ice cream is improbable, and levitating ice cream is impossible. But scooping ice cream using sadness is not just impossible: it is inconceivable. While prior work has examined the distinction between improbable and impossible events, there has been little empirical research on inconceivability. Here, we report a behavioral and computational study of inconceivability in three parts. First, we find that humans reliably categorize events as inconceivable, separate from probable, improbable, and impossible. Second, we find that we can decode the modal category of a sentence using language-model-derived estimates of subjective event probabilities. Third, we reproduce a recent finding that improbable events yield slowest response times in a possibility judgment task, and show that inconceivable events are faster to judge than impossible and improbable events. Overall, our results suggest that people distinguish the impossible from the inconceivable, and such distinctions may be based on graded rather than discrete judgments.

Sometimes people ask us to do things we do not want to do. We may try to avoid the aversive consequences of non-compliance by finding a loophole: an interpretation of the request that is consistent with its literal but not intended meaning. Exploiting loopholes requires an integrated understanding of pragmatics, goal alignment, and rational planning. This kind of complex social reasoning may be challenging for people with autism. Here we surveyed parents to study the prevalence and development of loophole behavior in childhood among autistic and non-autistic children. Neither the tendency to produce loopholes nor their developmental trajectory differed between autistic (N = 200) and non-autistic children (N = 200). These results are consistent with previous work suggesting the heterogeneous nature of autism and the difficulty of finding single tasks that distinguish high-functioning children with and without autism; the results also demonstrate that autistic children are capable of this kind of complex social reasoning.

Machine creativity is on the rise. Recent studies find that large language models achieve human performance on common psychological tests of creativity, which often pose a given problem and ask for novel or unusual solutions. But can AI go beyond producing solutions for given problems, to creatively propose new problems? We present the Task Task, a novel test that asks participants to come up with creative problems. In this test, we assess the ability of humans and GPT-4 to design challenge tasks for a game show. We evaluated proposed tasks using crowdsourced subjective creativity ratings, as well as computational measures of linguistic complexity and semantic content. We found that GPT-4 achieves similar scores as humans on creativity, originality, and judgments of how fun or difficult the tasks are. However, model-generated output tends to be shorter and connect more semantically distant concepts. We discuss implications and future directions for the psychology of creativity.

People engage in intentional misunderstandings to get around direct non-compliance. In other words, they use loopholes. Previous work showed that adults and children consider loophole behavior to be less costly than direct non-compliance (Bridgers, Schulz, & Ullman, 2021), and suggested this is a primary reason for their use: loopholes will land you in less trouble than defiance. However, we propose that this difference between loopholes and defiance will not hold for a specific, important context: moral violations. We replicate the finding that loopholes are less costly in a neutral context but find that engaging in loopholes in a moral context is as bad as non-compliance (Experiment 1, N=360). We then use a direct comparison between loopholes and non-compliance (N=150) to investigate whether in some contexts loopholes will be seen as even worse than non-compliance. We replicate the differential effect of the moral context from Experiment 1, but do not find a reversal. We discuss possible extensions and limitations, and consider why loopholes in moral violations may be uniquely unacceptable.

Decision-making can be automatic (“rote”) or on-the-fly (“reflective”). Are people sensitive to whether others are behaving rotely or reflectively? The rote-versus-reflective inference may be particularly relevant when learning from others: good teachers should be actively considering the learner’s needs. When teachers rely on rote systems, this may “break” the mental state recursion that facilitates learning from pedagogy. This study takes a first step in investigating learning and inference when teachers use rote-versus-reflective reasoning, using uniqueness of feedback as a cue to reflectivity. Adult participants viewed videos of teachers providing either identical, similar, or unique feedback to three different groups of students; participants evaluated that teacher along several metrics. Consistent with our predictions, rote teachers were evaluated as poorer informants than reflective teachers, and students paired with rote teachers were expected to learn less. These results are the first to demonstrate sensitivity to, and impact from, inferences about others’ rote-versus-reflective behaviors.

We examine whether people evaluate the performance of other agents using only behavioral metrics (intuitive behaviorism) or by also taking into account the program that is driving an agent’s performance (intuitive cognitivism). In an online study, 200 participants, most without programming experience, learned to use a simple block programming language that controls a maze-solving robot. Participants then evaluated which of two programs was “better”, for 18 pairs of robots and programs. The programs varied in 3 metrics. Two of these, action efficiency and representation efficiency, were motivated by work in Reinforcement Learning and philosophy of mind, and the third, semantic generalization, is novel. We found that people’s judgements are in line with intuitive cognitivism, that people are sensitive to the program features, and that people intuitively evaluate and care about which other problems a program solves beyond the given task.

When human adults make decisions (e.g., wearing a seat belt), we often consider the negative consequences that would ensue if our actions were to fail, even if we have never experienced such a failure. Do the same considerations guide our understanding of other people's decisions? In this paper, we investigated whether adults, who have many years of experience making such decisions, and 6- and 7-year-old children, who have less experience and are demonstrably worse at judging the consequences of their own actions, conceive others' actions as motivated both by reward (how good reaching one's intended goal would be), and by what we call "danger" (how badly one's action could end). In two pre-registered experiments, we tested whether adults and 6- and 7-year-old children tailor their predictions and explanations of an agent's action choices to the specific degree of danger and reward entailed by each action. Across four different tasks, we found that children and adults expected others to negatively appraise dangerous situations and minimize the danger of their actions. Children's and adults' judgments varied systematically in accord with both the degree of danger the agent faced and the value the agent placed on the goal state it aimed to achieve. However, children did not calibrate their inferences about how much an agent valued the goal state of a successful action in accord with the degree of danger the action entailed, and adults calibrated these inferences more weakly than inferences concerning the agent's future action choices. These results suggest that from childhood, people use a degree of danger and reward to make quantitative, fine-grained explanations and predictions about other people's behavior, consistent with computational models on theory of mind that contain continuous representations of other agents' action plans.

Finding and exploiting loopholes is a familiar facet of fable, law, and everyday life. But cognitive, computational, and empirical work on this behavior remains scarce. Engaging with loopholes requires a nuanced understanding of goals, social ambiguity, and value alignment. We trace loophole behavior to early childhood, and we propose that exploiting loopholes results from a conflict in actors' goals combined with a pressure to cooperate. A survey of 260 parents reporting on 425 children reveals that loophole behavior is prevalent, frequent, and diverse in daily parent-child interactions, emerging around ages five to six and tapering off from around ages nine to ten into adolescence. A further experiment shows that adults consider loophole behavior in children as less costly than non-compliance, and children increasingly differentiate loophole behavior from non-compliance from ages four to ten. We discuss limitations of the current work together with a proposal for a formal framework for loophole behavior.

Humans routinely make inferences about both the contents and the workings of other minds based on observed actions. People consider what others want or know, but also how intelligent, rational, or attentive they might be. Here, we introduce a new methodology for quantitatively studying the mechanisms people use to attribute intelligence to others based on their behavior. We focus on two key judgments previously proposed in the literature: judgments based on observed outcomes (you're smart if you won the game) and judgments based on evaluating the quality of an agent's planning that led to their outcomes (you're smart if you made the right choice, even if you didn't succeed). We present a novel task, the maze search task (MST), in which participants rate the intelligence of agents searching a maze for a hidden goal. We model outcome-based attributions based on the observed utility of the agent upon achieving a goal, with higher utilities indicating higher intelligence, and model planning-based attributions by measuring the proximity of the observed actions to an ideal planner, such that agents who produce closer approximations of optimal plans are seen as more intelligent. We examine human attributions of intelligence in three experiments that use MST and find that participants used both outcome and planning as indicators of intelligence. However, observing the outcome was not necessary, and participants still made planning-based attributions of intelligence when the outcome was not observed. We also found that the weights individuals placed on plans and on outcome correlated with an individual's ability to engage in cognitive reflection. Our results suggest that people attribute intelligence based on plans given sufficient context and cognitive resources and rely on the outcome when computational resources or context are limited.

When we reason about the goals of others, how do we balance the positive outcomes that actions led to, with the potentiallybad ways those actions could have ended? In a four-part experiment, we tested whether and how adults (full study) and6- to 8-year-old children (ongoing study) expect other agents to take account of the ways their goal-directed action couldhave failed. Across 4 different tasks, we found that adults expected others to negatively appraise perilous situations (deeptrenches), to minimize the danger of their actions, and to trade off danger and reward in their action plans. Our preliminarychildrens study shows similar trends. These results suggest that people appeal to peril-how badly things could go if onesactions fail-when explaining and predicting other peoples actions, and also make quantitative inferences that are finelytuned to the degree of peril and reward that others face.

People make fast, spontaneous, and consistent judgementsof social situations, even in complex physical contexts withmultiple-body dynamics (e.g. pushing, lifting, carrying, etc.).What mental computations make such judgments possible? Dopeople rely on low-level perceptual cues, or on abstract con-cepts of agency, action, and force? We describe a new exper-imental paradigm, Flatland, for studying social inference inphysical environments, using automatically generated interac-tive scenarios. We show that human interpretations of events inFlatland can be explained by a computational model that com-bines inverse hierarchical planning with a physical simulationengine to reason about objects and agents. This model out-performs cue-based alternatives based on hand-coded (multi-nomial logistic regression) and learned (LSTM) features. Ourresults suggest that humans could use a combination of intu-itive physics and hierarchical planning to interpret complex in-teractive scenarios encountered in daily life.

We are surprised when events violate our intuitive physicalexpectations. Even infants look longer when things seem tomagically teleport or vanish. This important surprise signalhas been used to probe what infants expect, in order to studythe most basic representations of objects. But these studiesrely on binary measures – an event is surprising, or not. Here,we study surprise in a more precise, quantitative way, usingthree distinct measures: we ask adults to judge how surprisinga scene is, when that scene is surprising, and why it is surpris-ing. We find good consistency in the level of surprise reportedacross these experiments, but also crucial differences in theimplied explanations of those scenes. Beyond this, we showthat the timing and degree of surprise can be explained by anobject-based model of intuitive physics.

The standard model of theory of mind posits that we attribute mental states to other people to explain their behavior. However, what of cases in which we think the other person is being scripted, acting automatically with no goals or beliefs to recover? While a great deal of past work has distinguished between automatic and reflective behaviors in one's own decision making, here we argue that reasoning about automatic behavior in other people is an important and largely unexplored area in research into theory of mind. We report results from two studies (N = 4,528 total) that examine the detection of automatic behavior in others. In Study 1, we conducted a large-scale survey characterizing the ubiquity of rote interactions in people's daily lives. In Study 2, we showed participants short video clips from a variety of domains and found that people quickly and reliably attribute automaticity to others and that automaticity judgments are distinct from other related behavioral attributions. On the basis of our findings, we suggest that reasoning about scripted behavior in others is an important, frequent, intuitive inference and propose extensions to the current research in intuitive psychology to study it further. Public Significance Statement: When interacting with another person, you may have the sense that the other person is acting in a scripted or automatic way. We find that such interactions are highly frequent and that the perception of automatic behavior is fast and consistent. We propose that research, which examines how we reason about the mental life of others, should be expanded to include reasoning about other people behaving automatically. [ABSTRACT FROM AUTHOR]

Humans have an intuitive understanding of physics. They canpredict how a physical scene will unfold, and reason about howit came to be. Adults may rely on such a physical representa-tion for visual reasoning and recognition, going beyond visualfeatures and capturing objects in terms of their physical prop-erties. Recently, the use of draped objects in recognition wasused to examine adult object representations in the absence ofmany common visual features. In this paper we examine youngchildren’s reasoning about draped objects in order to examinethe develop of physical object representation. In addition, weargue that a better understanding of the development of theconcept of cloth as a physical entity is worthwhile in and ofitself, as it may form a basic ontological category in intuitivephysical reasoning akin to liquids and solids. We use two ex-periments to investigate young children’s (ages 3–5) reasoningabout cloth-covered objects, and find that they perform signif-icantly above chance (though far from perfectly) indicating arepresentation of physical objects that can interact dynamicallywith the world. Children’s success and failure pattern is similaracross the two experiments, and we compare it to adult behav-ior. We find a small effect, which suggests the specific featuresthat make reasoning about certain objects more difficult maycarry into adulthood.

Constructing an intuitive theory from data confronts learners with a "chicken-and-egg" problem: The laws can only be expressed in terms of the theory's core concepts, but these concepts are only meaningful in terms of the role they play in the theory's laws; how can a learner discover appropriate concepts and laws simultaneously, knowing neither to begin with? We explore how children can solve this chicken-and-egg problem in the domain of magnetism, drawing on perspectives from computational modeling and behavioral experiments. We present 4- and 5-year-olds with two different simplified magnet-learning tasks. Children appropriately constrain their beliefs to two hypotheses following ambiguous but informative evidence. Following a critical intervention, they learn the correct theory. In the second study, children infer the correct number of categories given no information about the possible causal laws. Children's hypotheses in these tasks are explained as rational inferences within a Bayesian computational framework.

Languages commonly use physical properties to discuss dis-tinctly non-physical states and events in the world (e.g., “I’mnot a huge fan of licorice”). Here, we investigate the degreeto which associations between physical properties and abstractconcepts are culturally specific constructs. To do this, wetested three distinct populations—US adults, US children, andadults from an indigenous group in the lowlands of Bolivia, theTsimane’—on their associations between the physical conceptof weight and a variety of abstract attributes (e.g., importance,emotional state, moral worth). We find a strong relationshipbetween the associations of US and Tsimane’ adults, but little-to-no relationship between US children and either adult popu-lation. These results suggest that the property of weight playsa similar role in everyday thought across cultures, but that ittakes time to develop. Further, we found that these associationscould not be recovered from a simple semantic embeddinganalysis, suggesting that the cross-culturally shared connec-tions between physical and abstract attributes may be learnedthrough more complex experiences than language alone.

Models of concept learning and theory acquisition often in-voke a stochastic search process, in which learners generatehypotheses through some structured random process and thenevaluate them on some data measuring their quality or value.To be successful within a reasonable time-frame, these mod-els need ways of generating good candidate hypotheses evenbefore the data are considered. Schulz (2012a) has proposedthat studying the origins of new ideas in more everyday con-texts, such as how we think up new names for things, can pro-vide insight into the cognitive processes that generate good hy-potheses for learning. We propose a simple generative modelfor how people might draw on their experience to proposenew names in everyday domains such as pub names or actionmovies, and show that it captures surprisingly well the namesthat people actually imagine. We discuss the role for an anal-ogous hypothesis-generation mechanism in enabling and con-straining causal theory learning.

How do people understand the pragmatics of spatial language?We propose a rational-speech act model for spatial reasoning,and apply it to the terms ‘in’ and ‘near’. We examine people’sfine-grain spatial reasoning in this domain by having them lo-cate where an event occurred, given an utterance. Our prag-matic listener model provides a quantitative and qualitative fitto people’s inferences.

We present an algorithmic model for the development of children's intuitive theories within a hierarchical Bayesian framework, where theories are described as sets of logical laws generated by a probabilistic context-free grammar. We contrast our approach with connectionist and other emergentist approaches to modeling cognitive development. While their subsymbolic representations provide a smooth error surface that supports efficient gradient-based learning, our symbolic representations are better suited to capturing children's intuitive theories but give rise to a harder learning problem, which can only be solved by exploratory search. Our algorithm attempts to discover the theory that best explains a set of observed data by performing stochastic search at two levels of abstraction: an outer loop in the space of theories and an inner loop in the space of explanations or models generated by each theory given a particular dataset. We show that this stochastic search is capable of learning appropriate theories in several everyday domains and discuss its dynamics in the context of empirical studies of children's learning. (Contains 8 figures.)

The very early appearance of abstract knowledge is often taken as evidence for innateness. We explore the relative learning speeds of abstract and specific knowledge within a Bayesian framework and the role for innate structure. We focus on knowledge about causality, seen as a domain-general intuitive theory, and ask whether this knowledge can be learned from co-occurrence of events. We begin by phrasing the causal Bayes nets theory of causality and a range of alternatives in a logical language for relational theories. This allows us to explore simultaneous inductive learning of an abstract theory of causality and a causal model for each of several causal systems. We find that the correct theory of causality can be learned relatively quickly, often becoming available before specific causal theories have been learned--an effect we term the "blessing of abstraction". We then explore the effect of providing a variety of auxiliary evidence and find that a collection of simple perceptual input analyzers can help to bootstrap abstract knowledge. Together, these results suggest that the most efficient route to causal knowledge may be to build in not an abstract notion of causality but a powerful inductive learning mechanism and a variety of perceptual supports. While these results are purely computational, they have implications for cognitive development, which we explore in the conclusion. (Contains 1 table, 5 figures and 3 footnotes.)