Search

Establishing a unified theory of cognition has been a major goal of psychology. While there have been previous attempts to instantiate such theories by building computational models, we currently do not have one model that captures the human mind in its entirety. Here we introduce Centaur, a computational model that can predict and simulate human behavior in any experiment expressible in natural language. We derived Centaur by finetuning a state-of-the-art language model on a novel, large-scale data set called Psych-101. Psych-101 reaches an unprecedented scale, covering trial-by-trial data from over 60,000 participants performing over 10,000,000 choices in 160 experiments. Centaur not only captures the behavior of held-out participants better than existing cognitive models, but also generalizes to new cover stories, structural task modifications, and entirely new domains. Furthermore, we find that the model's internal representations become more aligned with human neural activity after finetuning. Taken together, Centaur is the first real candidate for a unified model of human cognition. We anticipate that it will have a disruptive impact on the cognitive sciences, challenging the existing paradigm for developing computational models.

Periorbital segmentation and distance prediction using deep learning allows for the objective quantification of disease state, treatment monitoring, and remote medicine. However, there are currently no reports of segmentation datasets for the purposes of training deep learning models with sub mm accuracy on the regions around the eyes. All images (n=2842) had the iris, sclera, lid, caruncle, and brow segmented by five trained annotators. Here, we validate this dataset through intra and intergrader reliability tests and show the utility of the data in training periorbital segmentation networks. All the annotations are publicly available for free download. Having access to segmentation datasets designed specifically for oculoplastic surgery will permit more rapid development of clinically useful segmentation networks which can be leveraged for periorbital distance prediction and disease classification. In addition to the annotations, we also provide an open-source toolkit for periorbital distance prediction from segmentation masks. The weights of all models have also been open-sourced and are publicly available for use by the community., Comment: 10 pages, 7 figures, 3 Tables

Periorbital distances and features around the eyes and lids hold valuable information for disease quantification and monitoring of surgical and medical intervention. These distances are commonly measured manually, a process that is both subjective and highly time-consuming. Here, we set out to developed three deep-learning methods for segmentation and periorbital distance prediction, and also evaluate the utility of periorbital distances for disease classification. The MAE of our deep learning predicted distances was less than or very close to the error observed between trained human annotators. We compared our models to the current state-of-the-art (SOTA) method for periorbital distance prediction and found that our methods outperformed SOTA on all of our datasets on all but one periorbital measurement. We also show that robust segmentation can be achieved on diseased eyes using models trained on open-source, healthy eyes, and that periorbital distances have can be used as high-quality features in downstream classification models. Leveraging segmentation networks as intermediary steps in classification has broad implications for increasing the generalizability of classification models in ophthalmic plastic and craniofacial surgery by avoiding the out-of-distribution problem observed in traditional convolutional neural networks., Comment: 20 pages, 10 figures, 9 tables

In order to solve problems of practical importance, quantum computers will likely need to incorporate quantum error correction, where a logical qubit is redundantly encoded in many noisy physical qubits. The large physical-qubit overhead typically associated with error correction motivates the search for more hardware-efficient approaches. Here, using a microfabricated superconducting quantum circuit, we realize a logical qubit memory formed from the concatenation of encoded bosonic cat qubits with an outer repetition code of distance $d=5$. The bosonic cat qubits are passively protected against bit flips using a stabilizing circuit. Cat-qubit phase-flip errors are corrected by the repetition code which uses ancilla transmons for syndrome measurement. We realize a noise-biased CX gate which ensures bit-flip error suppression is maintained during error correction. We study the performance and scaling of the logical qubit memory, finding that the phase-flip correcting repetition code operates below threshold, with logical phase-flip error decreasing with code distance from $d=3$ to $d=5$. Concurrently, the logical bit-flip error is suppressed with increasing cat-qubit mean photon number. The minimum measured logical error per cycle is on average $1.75(2)\%$ for the distance-3 code sections, and $1.65(3)\%$ for the longer distance-5 code, demonstrating the effectiveness of bit-flip error suppression throughout the error correction cycle. These results, where the intrinsic error suppression of the bosonic encodings allows us to use a hardware-efficient outer error correcting code, indicate that concatenated bosonic codes are a compelling paradigm for reaching fault-tolerant quantum computation., Comment: Comments on the manuscript welcome!

Understanding how people behave in strategic settings--where they make decisions based on their expectations about the behavior of others--is a long-standing problem in the behavioral sciences. We conduct the largest study to date of strategic decision-making in the context of initial play in two-player matrix games, analyzing over 90,000 human decisions across more than 2,400 procedurally generated games that span a much wider space than previous datasets. We show that a deep neural network trained on these data predicts people's choices better than leading theories of strategic behavior, indicating that there is systematic variation that is not explained by those theories. We then modify the network to produce a new, interpretable behavioral model, revealing what the original network learned about people: their ability to optimally respond and their capacity to reason about others are dependent on the complexity of individual games. This context-dependence is critical in explaining deviations from the rational Nash equilibrium, response times, and uncertainty in strategic decisions. More broadly, our results demonstrate how machine learning can be applied beyond prediction to further help generate novel explanations of complex human behavior.

In order for AI systems to communicate effectively with people, they must understand how we make decisions. However, people's decisions are not always rational, so the implicit internal models of human decision-making in Large Language Models (LLMs) must account for this. Previous empirical evidence seems to suggest that these implicit models are accurate -- LLMs offer believable proxies of human behavior, acting how we expect humans would in everyday interactions. However, by comparing LLM behavior and predictions to a large dataset of human decisions, we find that this is actually not the case: when both simulating and predicting people's choices, a suite of cutting-edge LLMs (GPT-4o & 4-Turbo, Llama-3-8B & 70B, Claude 3 Opus) assume that people are more rational than we really are. Specifically, these models deviate from human behavior and align more closely with a classic model of rational choice -- expected value theory. Interestingly, people also tend to assume that other people are rational when interpreting their behavior. As a consequence, when we compare the inferences that LLMs and people draw from the decisions of others using another psychological dataset, we find that these inferences are highly correlated. Thus, the implicit decision-making models of LLMs appear to be aligned with the human expectation that other people will act rationally, rather than with how people actually act.

Running quantum algorithms protected by quantum error correction requires a real time, classical decoder. To prevent the accumulation of a backlog, this decoder must process syndromes from the quantum device at a faster rate than they are generated. Most prior work on real time decoding has focused on an isolated logical qubit encoded in the surface code. However, for surface code, quantum programs of utility will require multi-qubit interactions performed via lattice surgery. A large merged patch can arise during lattice surgery -- possibly as large as the entire device. This puts a significant strain on a real time decoder, which must decode errors on this merged patch and maintain the level of fault-tolerance that it achieves on isolated logical qubits. These requirements are relaxed by using spatially parallel decoding, which can be accomplished by dividing the physical qubits on the device into multiple overlapping groups and assigning a decoder module to each. We refer to this approach as spatially parallel windows. While previous work has explored similar ideas, none have addressed system-specific considerations pertinent to the task or the constraints from using hardware accelerators. In this work, we demonstrate how to configure spatially parallel windows, so that the scheme (1) is compatible with hardware accelerators, (2) supports general lattice surgery operations, (3) maintains the fidelity of the logical qubits, and (4) meets the throughput requirement for real time decoding. Furthermore, our results reveal the importance of optimally choosing the buffer width to achieve a balance between accuracy and throughput -- a decision that should be influenced by the device's physical noise.

Quantum error correction with erasure qubits promises significant advantages over standard error correction due to favorable thresholds for erasure errors. To realize this advantage in practice requires a qubit for which nearly all errors are such erasure errors, and the ability to check for erasure errors without dephasing the qubit. We demonstrate that a "dual-rail qubit" consisting of a pair of resonantly coupled transmons can form a highly coherent erasure qubit, where transmon $T_1$ errors are converted into erasure errors and residual dephasing is strongly suppressed, leading to millisecond-scale coherence within the qubit subspace. We show that single-qubit gates are limited primarily by erasure errors, with erasure probability $p_\text{erasure} = 2.19(2)\times 10^{-3}$ per gate while the residual errors are $\sim 40$ times lower. We further demonstrate mid-circuit detection of erasure errors while introducing $< 0.1\%$ dephasing error per check. Finally, we show that the suppression of transmon noise allows this dual-rail qubit to preserve high coherence over a broad tunable operating range, offering an improved capacity to avoid frequency collisions. This work establishes transmon-based dual-rail qubits as an attractive building block for hardware-efficient quantum error correction., Comment: 9+13 pages, 16 figures

Shepard's universal law of generalization is a remarkable hypothesis about how intelligent organisms should perceive similarity. In its broadest form, the universal law states that the level of perceived similarity between a pair of stimuli should decay as a concave function of their distance when embedded in an appropriate psychological space. While extensively studied, evidence in support of the universal law has relied on low-dimensional stimuli and small stimulus sets that are very different from their real-world counterparts. This is largely because pairwise comparisons -- as required for similarity judgments -- scale quadratically in the number of stimuli. We provide direct evidence for the universal law in a naturalistic high-dimensional regime by analyzing an existing dataset of 214,200 human similarity judgments and a newly collected dataset of 390,819 human generalization judgments (N=2406 US participants) across three sets of natural images., Comment: 36 pages, 6 figures

Supervised learning typically focuses on learning transferable representations from training examples annotated by humans. While rich annotations (like soft labels) carry more information than sparse annotations (like hard labels), they are also more expensive to collect. For example, while hard labels only provide information about the closest class an object belongs to (e.g., "this is a dog"), soft labels provide information about the object's relationship with multiple classes (e.g., "this is most likely a dog, but it could also be a wolf or a coyote"). We use information theory to compare how a number of commonly-used supervision signals contribute to representation-learning performance, as well as how their capacity is affected by factors such as the number of labels, classes, dimensions, and noise. Our framework provides theoretical justification for using hard labels in the big-data regime, but richer supervision signals for few-shot learning and out-of-distribution generalization. We validate these results empirically in a series of experiments with over 1 million crowdsourced image annotations and conduct a cost-benefit analysis to establish a tradeoff curve that enables users to optimize the cost of supervising representation learning on their own datasets., Comment: Proceedings of UAI 2023

We describe a distributed, asynchronous variant of Edmonds's exact algorithm for producing perfect matchings of minimum weight. The development of this algorithm is driven by an application to online error correction in quantum computing, first envisioned by Fowler; we analyze the performance of our algorithm as applied to this domain in a sequel.

We demonstrate how to determine numerically nearly exact orthonormal orbitals that are optimal for evaluation of the energy of arbitrary (correlated) states of atoms and molecules by minimization of the energy Lagrangian. Orbitals are expressed in real space using a multiresolution spectral element basis that is refined adaptively to achieve the user-specified target precision while avoiding the ill-conditioning issues that plague AO basis set expansions traditionally used for correlated models of molecular electronic structure. For light atoms, the orbital solver, in conjunction with a variational electronic structure model [selected Configuration Interaction(CI)] provides energies of comparable precision to a state-of-the-art atomic CI solver. The computed electronic energies of atoms and molecules are significantly more accurate than the counterparts obtained with the Gaussian AO bases of the same rank, and can be determined even when linear dependence issues preclude the use of the AO bases. It is feasible to optimize more than 100 fully-correlated numerical orbitals on a single computer node, and significant room exists for additional improvement. These findings suggest that the real-space orbital representations might be the preferred alternative to AO representations for high-end models of correlated electronic states of molecules and materials.

The spectrum of neutral iron is critical to astrophysics, yet furnace laboratory experiments cannot reach many high-lying Fe I levels. Instead, Peterson & Kurucz (2015) and Peterson, Kurucz & Ayres (2017) turned to UV and optical spectra of warm stars to identify and assign energies for 124 Fe I levels with 1900 detectable Fe I lines, and to derive astrophysical gf values for over a thousand of these. An energy value was assumed for each unknown Fe I level, and confirmed if it shifted the predicted positions in updated Kurucz (2011) Fe I calculations to match exactly in wavelength the positions of four or more unidentified lines in the observed spectra. Nearly all these identifications were for LS levels characterized by spin-orbit coupling, whose lines fall primarily at UV and optical wavelengths. This extension of these searches provides nearly a hundred new Fe I level identifications. Thirty-nine LS levels are identified largely by incorporating published positions of unidentified laboratory Fe I lines with wavelengths < 2000A. Adding infrared spectra provided sixty Fe I jK levels, where a single isolated outer electron orbits a compact core. Their weak, blended lines fall mostly in the infrared, but are searchable because their mutual energies obey tight relationships. For each new Fe I level, this work again provides and makes publicly available its identification, its energy, and a list of over 16,000 of its potentially detectable lines with theoretical gf values. For over 2,000 of these, this work also includes astrophysical gf values, ones adjusted semi-empirically to fit the stellar spectra. The potential impact of this work on modeling UV and IR stellar spectra is noted., Comment: 23 pages, 1 figure. Near-final ApJS version. Section 9 characterizes and quantifies uncertainties in deduced level energies and astrophysical dgf values. Table 3 includes >16,000 Fe I lines potentially detectable in solar-metallicity dwarfs and giants, and dgf values for >2,000 of these. It is currently available from the website of R. Kurucz at http://kurucz.harvard.edu/atoms/2600/PK21/table3.txt

We describe an optimal procedure, as well as its efficient software implementation, for exact and approximate synthesis of two-qubit unitary operations into any prescribed discrete family of XX-type interactions and local gates. This arises from the analysis and manipulation of certain polyhedral subsets of the space of canonical gates. Using this, we analyze which small sets of XX-type interactions cause the greatest improvement in expected infidelity under experimentally-motivated error models. For the exact circuit synthesis of Haar-randomly selected two-qubit operations, we find an improvement in estimated infidelity by ~31.4% when including alongside CX its square- and cube-roots, near to the optimal limit of ~36.9% obtained by including all fractional applications of CX., Comment: Camera-ready version

Cowan et al. (2021) review how roughly half the elements heavier than iron found in the Sun are produced by rapid neutron capture and half by slow neutron capture, the r- and s-processes. In the Sun, their relative contribution to individual elemental abundances is well understood, except for the lightest and heaviest elements beyond iron. Their contributions are especially uncertain for the heaviest non-radioactive element, lead (Pb, Z=82). This is constrained by deriving lead abundances in metal-poor stars. For in the most metal-poor halo stars, strontium and heavier elements are found in the solar r-process proportion; s-process elements appear only at metallicities above one-thirtieth solar. In unevolved metal-poor stars of roughly solar heavy-element content, only two UV Pb lines are detectable. Four such stars have high-resolution spectra of the strongest line, Pb II at 2203.53A. Roederer et al. (2020) analyzed this line in one star, deriving a lead-to-iron abundance ratio ten times solar. This and its blue-shifted profile suggested strong s-process production. This work analyzes the UV spectra of all four stars. Calculations including a predicted Fe I line blueward of the Pb II line, and assuming the lead abundance scales with r-process abundances, match all four profiles extremely well. A scaled s-process contribution might improve the match to the much lower lead abundance found in the unevolved star analyzed previously, but its s-process excess is modest. An Fe II line blends the other lead line, Pb I at 2833.05A, which constrains the lead abundance only in the coolest star., Comment: 9 pages, 3 figures. Accepted for publication in the Astrophysical Journal Letters as a gold open access article under a CC BY license

We describe a Common Lisp package suitable for the high-level design, specification, simulation, and instrumentation of real-time distributed algorithms and hardware on which to run them. We discuss various design decisions around the package structure, and we explore their consequences with small examples., Comment: Final, for-publication version

Traditional models of category learning in psychology focus on representation at the category level as opposed to the stimulus level, even though the two are likely to interact. The stimulus representations employed in such models are either hand-designed by the experimenter, inferred circuitously from human judgments, or borrowed from pretrained deep neural networks that are themselves competing models of category learning. In this work, we extend classic prototype and exemplar models to learn both stimulus and category representations jointly from raw input. This new class of models can be parameterized by deep neural networks (DNN) and trained end-to-end. Following their namesakes, we refer to them as Deep Prototype Models, Deep Exemplar Models, and Deep Gaussian Mixture Models. Compared to typical DNNs, we find that their cognitively inspired counterparts both provide better intrinsic fit to human behavior and improve ground-truth classification., Comment: 7 pages, 4 figures, 2 tables. Accepted as a paper to the 42nd Annual Meeting of the Cognitive Science Society (CogSci 2020)

Spectra of unevolved metal-poor halo stars uniquely reflect the elemental abundances incorporated during the earliest Galactic epoch. Their heavy-element content is well understood as the products of neutron capture on iron-peak elements. However, the lightest elements just past the iron peak, with atomic number 30

Quilc is an open-source, optimizing compiler for gate-based quantum programs written in Quil or QASM, two popular quantum programming languages. The compiler was designed with attention toward NISQ-era quantum computers, specifically recognizing that each quantum gate has a non-negligible and often irrecoverable cost toward a program's successful execution. Quilc's primary goal is to make authoring quantum software a simpler exercise by making architectural details less burdensome to the author. Using Quilc allows one to write programs faster while usually not compromising---and indeed sometimes improving---their execution fidelity on a given hardware architecture. In this paper, we describe many of the principles behind Quilc's design, and demonstrate the compiler with various examples.