Abstract:There has been recent interest in whether large language models (LLMs) can introspect about their own internal states. Such abilities would make LLMs more interpretable, and also validate the use of standard introspective methods in linguistics to evaluate grammatical knowledge in models (e.g., asking "Is this sentence grammatical?"). We systematically investigate emergent introspection across 21 open-source LLMs, in two domains where introspection is of theoretical interest: grammatical knowledge and word prediction. Crucially, in both domains, a model's internal linguistic knowledge can be theoretically grounded in direct measurements of string probability. We then evaluate whether models' responses to metalinguistic prompts faithfully reflect their internal knowledge. We propose a new measure of introspection: the degree to which a model's prompted responses predict its own string probabilities, beyond what would be predicted by another model with nearly identical internal knowledge. While both metalinguistic prompting and probability comparisons lead to high task accuracy, we do not find evidence that LLMs have privileged "self-access". Our findings complicate recent results suggesting that models can introspect, and add new evidence to the argument that prompted responses should not be conflated with models' linguistic generalizations.
Abstract:The question of whether large language models (LLMs) possess Theory of Mind (ToM) -- often defined as the ability to reason about others' mental states -- has sparked significant scientific and public interest. However, the evidence as to whether LLMs possess ToM is mixed, and the recent growth in evaluations has not resulted in a convergence. Here, we take inspiration from cognitive science to re-evaluate the state of ToM evaluation in LLMs. We argue that a major reason for the disagreement on whether LLMs have ToM is a lack of clarity on whether models should be expected to match human behaviors, or the computations underlying those behaviors. We also highlight ways in which current evaluations may be deviating from "pure" measurements of ToM abilities, which also contributes to the confusion. We conclude by discussing several directions for future research, including the relationship between ToM and pragmatic communication, which could advance our understanding of artificial systems as well as human cognition.
Abstract:Some things are impossible, but some things may be even more impossible than impossible. Levitating a feather using one's mind is impossible in our world, but fits into our intuitive theories of possible worlds, whereas levitating a feather using the number five cannot be conceived in any possible world ("inconceivable"). While prior work has examined the distinction between improbable and impossible events, there has been little empirical research on inconceivability. Here, we investigate whether people maintain a distinction between impossibility and inconceivability, and how such distinctions might be made. We find that people can readily distinguish the impossible from the inconceivable, using categorization studies similar to those used to investigate the differences between impossible and improbable (Experiment 1). However, this distinction is not explained by people's subjective ratings of event likelihood, which are near zero and indistinguishable between impossible and inconceivable event descriptions (Experiment 2). Finally, we ask whether the probabilities assigned to event descriptions by statistical language models (LMs) can be used to separate modal categories, and whether these probabilities align with people's ratings (Experiment 3). We find high-level similarities between people and LMs: both distinguish among impossible and inconceivable event descriptions, and LM-derived string probabilities predict people's ratings of event likelihood across modal categories. Our findings suggest that fine-grained knowledge about exceedingly rare events (i.e., the impossible and inconceivable) may be learned via statistical learning over linguistic forms, yet leave open the question of whether people represent the distinction between impossible and inconceivable as a difference not of degree, but of kind.
Abstract:Researchers in social science and psychology have recently proposed using large language models (LLMs) as replacements for humans in behavioral research. In addition to arguments about whether LLMs accurately capture population-level patterns, this has raised questions about whether LLMs capture human-like conceptual diversity. Separately, it is debated whether post-training alignment (RLHF or RLAIF) affects models' internal diversity. Inspired by human studies, we use a new way of measuring the conceptual diversity of synthetically-generated LLM "populations" by relating the internal variability of simulated individuals to the population-level variability. We use this approach to evaluate non-aligned and aligned LLMs on two domains with rich human behavioral data. While no model reaches human-like diversity, aligned models generally display less diversity than their instruction fine-tuned counterparts. Our findings highlight potential trade-offs between increasing models' value alignment and decreasing the diversity of their conceptual representations.
Abstract:The ability to build and leverage world models is essential for a general-purpose AI agent. Testing such capabilities is hard, in part because the building blocks of world models are ill-defined. We present Elements of World Knowledge (EWOK), a framework for evaluating world modeling in language models by testing their ability to use knowledge of a concept to match a target text with a plausible/implausible context. EWOK targets specific concepts from multiple knowledge domains known to be vital for world modeling in humans. Domains range from social interactions (help/hinder) to spatial relations (left/right). Both, contexts and targets are minimal pairs. Objects, agents, and locations in the items can be flexibly filled in enabling easy generation of multiple controlled datasets. We then introduce EWOK-CORE-1.0, a dataset of 4,374 items covering 11 world knowledge domains. We evaluate 20 openweights large language models (1.3B--70B parameters) across a battery of evaluation paradigms along with a human norming study comprising 12,480 measurements. The overall performance of all tested models is worse than human performance, with results varying drastically across domains. These data highlight simple cases where even large models fail and present rich avenues for targeted research on LLM world modeling capabilities.
Abstract:Developmental psychologists have argued about when cognitive capacities such as language understanding or theory of mind emerge. These debates often hinge on the concept of "task demands" -- the auxiliary challenges associated with performing a particular evaluation -- that may mask the child's underlying ability. The same issues arise when measuring the capacities of language models (LMs): performance on a task is a function of the model's underlying competence, combined with the model's ability to interpret and perform the task given its available resources. Here, we show that for analogical reasoning, reflective reasoning, word prediction, and grammaticality judgments, evaluation methods with greater task demands yield lower performance than evaluations with reduced demands. This "demand gap" is most pronounced for models with fewer parameters and less training data. Our results illustrate that LM performance should not be interpreted as a direct indication of intelligence (or lack thereof), but as a reflection of capacities seen through the lens of researchers' design choices.
Abstract:Do Large Language Models (LLMs) make human-like linguistic generalizations? Dentella et al. (2023; "DGL") prompt several LLMs ("Is the following sentence grammatically correct in English?") to elicit grammaticality judgments of 80 English sentences, concluding that LLMs demonstrate a "yes-response bias" and a "failure to distinguish grammatical from ungrammatical sentences". We re-evaluate LLM performance using well-established practices and find that DGL's data in fact provide evidence for just how well LLMs capture human behaviors. Models not only achieve high accuracy overall, but also capture fine-grained variation in human linguistic judgments.
Abstract:Prompting is now a dominant method for evaluating the linguistic knowledge of large language models (LLMs). While other methods directly read out models' probability distributions over strings, prompting requires models to access this internal information by processing linguistic input, thereby implicitly testing a new type of emergent ability: metalinguistic judgment. In this study, we compare metalinguistic prompting and direct probability measurements as ways of measuring models' knowledge of English. Broadly, we find that LLMs' metalinguistic judgments are inferior to quantities directly derived from representations. Furthermore, consistency gets worse as the prompt diverges from direct measurements of next-word probabilities. Our findings suggest that negative results relying on metalinguistic prompts cannot be taken as conclusive evidence that an LLM lacks a particular linguistic competence. Our results also highlight the lost value with the move to closed APIs where access to probability distributions is limited.
Abstract:Scalar inferences (SI) are a signature example of how humans interpret language based on unspoken alternatives. While empirical studies have demonstrated that human SI rates are highly variable -- both within instances of a single scale, and across different scales -- there have been few proposals that quantitatively explain both cross- and within-scale variation. Furthermore, while it is generally assumed that SIs arise through reasoning about unspoken alternatives, it remains debated whether humans reason about alternatives as linguistic forms, or at the level of concepts. Here, we test a shared mechanism explaining SI rates within and across scales: context-driven expectations about the unspoken alternatives. Using neural language models to approximate human predictive distributions, we find that SI rates are captured by the expectedness of the strong scalemate as an alternative. Crucially, however, expectedness robustly predicts cross-scale variation only under a meaning-based view of alternatives. Our results suggest that pragmatic inferences arise from context-driven expectations over alternatives, and these expectations operate at the level of concepts.
Abstract:We propose a novel task, G4C (Goal-driven Guidance Generation in Grounded Communication), for studying goal-driven and grounded natural language interactions. Specifically, we choose Dungeons and Dragons (D&D) -- a role-playing game consisting of multiple player characters and a Dungeon Master (DM) who collaborate to achieve a set of goals that are beneficial to the players -- as a testbed for this task. Here, each of the player characters is a student, with their own personas and abilities, and the DM is the teacher, an arbitrator of the rules of the world and responsible for assisting and guiding the students towards a global goal. We propose a theory-of-mind-inspired methodology for training such a DM with reinforcement learning (RL), where a DM: (1) learns to predict how the players will react to its utterances using a dataset of D&D dialogue transcripts; and (2) uses this prediction as a reward function providing feedback on how effective these utterances are at guiding the players towards a goal. Human and automated evaluations show that a DM trained with RL to generate guidance by incorporating a theory-of-mind of the players significantly improves the players' ability to achieve goals grounded in their shared world.