Insight
is our reward
Search

Publications in Information & Communication Technologies by NOMIS researchers

Binary climate data visuals amplify perceived impact of climate change

NOMIS Researcher(s)

Published in

April 17, 2025

For much of the global population, climate change appears as a slow, gradual shift in daily weather. This leads many to perceive its impacts as minor and results in apathy (the ‘boiling frog’ effect). How can we convey the urgency of the crisis when its impacts appear so subtle? Here, through a series of large-scale cognitive experiments (N = 799), we find that presenting people with binary climate data (for example, lake freeze history) significantly increases the perceived impact of climate change (Cohen’s d = 0.40, 95% confidence interval 0.26–0.54) compared with continuous data (for example, mean temperature). Computational modelling and follow-up experiments (N = 398) suggest that binary data enhance perceived impact by creating an ‘illusion’ of sudden shifts. Crucially, our approach does not involve selective data presentation but rather compares different datasets that reflect equivalent trends in climate change over time. These findings, robustly replicated across multiple experiments, provide a cognitive basis for the ‘boiling frog’ effect and offer a psychologically grounded approach for policymakers and educators to improve climate change communication while maintaining scientific accuracy.

Research field(s)
Information & Communication Technologies, Psychology & Cognitive Sciences, Behavioral Science & Comparative Psychology

DOI
10.1038/s41562-025-02183-9

AI-based approach to dissect the variability of mouse stem cell-derived embryo models

NOMIS Researcher(s)

Published in

February 19, 2025

Recent advances in stem cell-derived embryo models have transformed developmental biology, offering insights into embryogenesis without the constraints of natural embryos. However, variability in their development challenges research standardization. To address this, we use deep learning to enhance the reproducibility of selecting stem cell-derived embryo models. Through live imaging and AI-based models, we classify 900 mouse post-implantation stem cell-derived embryo-like structures (ETiX-embryos) into normal and abnormal categories. Our best-performing model achieves 88% accuracy at 90 h post-cell seeding and 65% accuracy at the initial cell-seeding stage, forecasting developmental trajectories. Our analysis reveals that normally developed ETiX-embryos have higher cell counts and distinct morphological features such as larger size and more compact shape. Perturbation experiments increasing initial cell numbers further supported this finding by improving normal development outcomes. This study demonstrates deep learning’s utility in improving embryo model selection and reveals critical features of ETiX-embryo self-organization, advancing consistency in this evolving field.

Research field(s)
Bioinformatics, Artificial Intelligence & Image Processing, Biophysics, Developmental Biology

DOI
10.1038/s41467-025-56908-5

Learning human-like representations to enable learning human values

NOMIS Researcher(s)

Published in

November 8, 2024

How can we build AI systems that can learn any set of individual human values both quickly and safely, avoiding causing harm or violating societal standards for acceptable behavior during the learning process? We explore the effects of representational alignment between humans and AI agents on learning human values. Making AI systems learn human-like representations of the world has many known benefits, including improving generalization, robustness to domain shifts, and few-shot learning performance. We demonstrate that this kind of representational alignment can also support safely learning and exploring human values in the context of personalization. We begin with a theoretical prediction, show that it applies to learning human morality judgments, then show that our results generalize to ten different aspects of human values — including ethics, honesty, and fairness — training AI agents on each set of values in a multi-armed bandit setting, where rewards reflect human value judgments over the chosen action. Using a set of textual action descriptions, we collect value judgments from humans, as well as similarity judgments from both humans and multiple language models, and demonstrate that representational alignment enables both safe exploration and improved generalization when learning human values.

Research field(s)
Artificial Intelligence & Image Processing, Psychology & Cognitive Sciences

DOI
10.48550/arXiv.2312.14106

Bayes in the age of intelligent machines

NOMIS Researcher(s)

Published in

September 21, 2024

The success of methods based on artificial neural networks in creating intelligent machines seems like it might pose a challenge to explanations of human cognition in terms of Bayesian inference. We argue that this is not the case and that these systems in fact offer new opportunities for Bayesian modeling. Specifically, we argue that artificial neural networks and Bayesian models of cognition lie at different levels of analysis and are complementary modeling approaches, together offering a way to understand human cognition that spans these levels. We also argue that the same perspective can be applied to intelligent machines, in which a Bayesian approach may be uniquely valuable in understanding the behavior of large, opaque artificial neural networks that are trained on proprietary data.

Research field(s)
Artificial Intelligence & Image Processing, Psychology & Cognitive Sciences

DOI
10.1177/09637214241262329

Learning from aggregated opinion

NOMIS Researcher(s)

Published in

July 24, 2024

The capacity to leverage information from others’ opinions is a hallmark of human cognition. Consequently, past research has investigated how we learn from others’ testimony. Yet a distinct form of social information—aggregated opinion—increasingly guides our judgments and decisions. We investigated how people learn from such information by conducting three experiments with participants recruited online within the United States (N = 886) comparing the predictions of three computational models: a Bayesian solution to this problem that can be implemented by a simple strategy for combining proportions with prior beliefs, and two alternatives from epistemology and economics. Across all studies, we found the strongest concordance between participants’ judgments and the predictions of the Bayesian model, though some participants’ judgments were better captured by alternative strategies. These findings lay the groundwork for future research and show that people draw systematic inferences from aggregated opinion, often in line with a Bayesian solution.

Research field(s)
Artificial Intelligence & Image Processing, Psychology & Cognitive Sciences

DOI
10.1177/09567976241251741

New Frontiers: The Origins and Content of New Work, 1940–2018

NOMIS Researcher(s)

Published in

March 15, 2024

We answer three core questions about the hypothesized role of newly emerging job categories (“new work”) in counterbalancing the erosive effect of task-displacing automation on labor demand: what is the substantive content of new work, where does it come from, and what effect does it have on labor demand? We construct a novel database spanning eight decades of new job titles linked to U.S. Census microdata and to patent-based measures of occupations’ exposure to labor-augmenting and labor-automating innovations. The majority of current employment is in new job specialties introduced since 1940, but the locus of new-work creation has shifted from middle-paid production and clerical occupations over 1940–1980 to high-paid professional occupations and secondarily to low-paid services since 1980. New work emerges in response to technological innovations that complement the outputs of occupations and demand shocks that raise occupational demand. Innovations that automate tasks or reduce occupational demand slow new-work emergence. Although the flow of augmentation and automation innovations is positively correlated across occupations, the former boosts occupational labor demand while the latter depresses it. The demand-eroding effects of automation innovations have intensified in the past four decades while the demand-increasing effects of augmentation innovations have not.

Research field(s)
Information & Communication Technologies, Economics

DOI
10.1093/qje/qjae008

Program-Based Strategy Induction for Reinforcement Learning

NOMIS Researcher(s)

Published in

February 26, 2024

Typical models of learning assume incremental estimation of continuously-varying decision variables like expected rewards. However, this class of models fails to capture more idiosyncratic, discrete heuristics and strategies that people and animals appear to exhibit. Despite recent advances in strategy discovery using tools like recurrent networks that generalize the classic models, the resulting strategies are often onerous to interpret, making connections to cognition difficult to establish. We use Bayesian program induction to discover strategies implemented by programs, letting the simplicity of strategies trade off against their effectiveness. Focusing on bandit tasks, we find strategies that are difficult or unexpected with classical incremental learning, like asymmetric learning from rewarded and unrewarded trials, adaptive horizon-dependent random exploration, and discrete state switching.

Research field(s)
Artificial Intelligence & Image Processing, Psychology & Cognitive Sciences

DOI
10.48550/arXiv.2402.16668

Incoherent probability judgments in large language models

NOMIS Researcher(s)

Published in

January 30, 2024

Autoregressive Large Language Models (LLMs) trained for next-word prediction have demonstrated remarkable proficiency at producing coherent text. But are they equally adept at forming coherent probability judgments? We use probabilistic identities and repeated judgments to assess the coherence of probability judgments made by LLMs. Our results show that the judgments produced by these models are often incoherent, displaying human-like systematic deviations from the rules of probability theory. Moreover, when prompted to judge the same event, the mean-variance relationship of probability judgments produced by LLMs shows an inverted-U-shaped like that seen in humans. We propose that these deviations from rationality can be explained by linking autoregressive LLMs to implicit Bayesian inference and drawing parallels with the Bayesian Sampler model of human probability judgments.

Research field(s)
Artificial Intelligence & Image Processing, Psychology & Cognitive Sciences

DOI
10.48550/arXiv.2401.16646

Machine learning for modeling human decisions

NOMIS Researcher(s)

Published in

January 1, 2024

The rapid development of machine learning has led to new opportunities for applying these methods to the study of human decision making. We highlight some of these opportunities and discuss some of the issues that arise when using machine learning to model the decisions people make. We first elaborate on the relationship between predicting decisions and explaining them, leveraging findings from computational learning theory to argue that, in some cases, the conversion of predictive models to interpretable ones with comparable accuracy is an intractable problem. We then identify an important bottleneck in using machine learning to study human cognition—data scarcity—and highlight active learning and optimal experimental design as a way to move forward. Finally, we touch on additional topics such as machine learning methods for combining multiple predictors arising from known theories and specific machine learning architectures that could prove useful for the study of judgment and decision making. In doing so, we point out connections to behavioral economics, computer science, cognitive science, and psychology. (PsycInfo Database Record (c) 2024 APA, all rights reserved)

Research field(s)
Artificial Intelligence & Image Processing, Psychology & Cognitive Sciences

DOI
10.1037/dec0000242

New approaches to epidemic modeling on networks

NOMIS Researcher(s)

Published in

January 10, 2023

In this article, we develop two independent and new approaches to model epidemic spread in a network. Contrary to the most studied models, those developed here allow for contacts with different probabilities of transmitting the disease (transmissibilities). We then examine each of these models using some mean field type approximations. The first model looks at the late-stage effects of an epidemic outbreak and allows for the computation of the probability that a given vertex was infected. This computation is based on a mean field approximation and only depends on the number of contacts and their transmissibilities. This approach shares many similarities with percolation models in networks. The second model we develop is a dynamic model which we analyze using a mean field approximation which highly reduces the dimensionality of the system. In particular, the original system which individually analyses each vertex of the network is reduced to one with as many equations as different transmissibilities. Perhaps the greatest contribution of this article is the observation that, in both these models, the existence and size of an epidemic outbreak are linked to the properties of a matrix which we call the R-matrix. This is a generalization of the basic reproduction number which more precisely characterizes the main routes of infection. © 2023, The Author(s).

Research field(s)
Applied Sciences, Information & Communication Technologies, Artificial Intelligence & Image Processing

DOI
10.1038/s41598-022-19827-9

Social Metacognition Drives Willingness to Commit

NOMIS Researcher(s)

Published in

January 1, 2023

Showing or telling others that we are committed to cooperate with them can boost social cooperation. But what makes us willing to signal our cooperativeness, when it is costly to do so? In two experiments,we tested the hypothesis that agents engage in social commitments if their subjective confidence in predicting the interaction partner’s behavior is low. In Experiment 1 (preregistered), 48 participants played a prisoner’s dilemma game where they could signal their intentions to their co-player by enduring a monetary cost. As hypothesized, low confidence in one’s prediction of the co-player’s intentions was associated with a higher willingness to engage in costly commitment. In Experiment 2 (31 participants), we replicate these findings and moreover provide causal evidence that experimentally lowering the predictability of others’ actions (and thereby confidence in these predictions) motivates commitment decisions. Finally, across both experiments, we show that participants possess and demonstrate metacognitive access to the accuracy of their mentalizing processes. Taken together, our findings shed light on the importance of confidence representations and metacognitive processes in social interactions © 2023 American Psychological Association

Research field(s)
Applied Sciences, Information & Communication Technologies, Artificial Intelligence & Image Processing

DOI
10.1037/xge0001419

Microwave quantum illumination with a digital phase-conjugated receiver

NOMIS Researcher(s)

Published in

September 21, 2020

Quantum illumination is a sensing technique that employs entangled signal-idler beams to improve the detection efficiency of low-reflectivity objects in environments with large thermal noise. The advantage over classical strategies is evident at low signal brightness, a feature which could make the protocol an ideal prototype for non-invasive scanning or low-power short-range radar. Here we experimentally investigate the concept of quantum illumination at microwave frequencies, by generating entangled fields using a Josephson parametric converter which are then amplified to illuminate a room-temperature object at a distance of 1 meter. Starting from experimental data, we simulate the case of perfect idler photon number detection, which results in a quantum advantage compared to the relative classical benchmark. Our results highlight the opportunities and challenges on the way towards a first room-temperature application of microwave quantum circuits.

Research field(s)
Applied Sciences, Information & Communication Technologies, Networking & Telecommunications

DOI
10.1109/RadarConf2043947.2020.9266397

Self-serving incentives impair collective decisions by increasing conformity

NOMIS Researcher(s)

Published in

November 1, 2019

The average judgment of large numbers of people has been found to be consistently better than the best individual response. But what motivates individuals when they make collective decisions? While it is a popular belief that individual incentives promote out-of-the-box thinking and diverse solutions, the exact role of motivation and reward in collective intelligence remains unclear. Here we examined collective intelligence in an interactive group estimation task where participants were rewarded for their individual or group’s performance. In addition to examining individual versus collective incentive structures, we controlled whether participants could see social information about the others’ responses. We found that knowledge about others’ responses reduced the wisdom of the crowd and, crucially, this effect depended on how people were rewarded. When rewarded for the accuracy of their individual responses, participants converged to the group mean, increasing social conformity, reducing diversity and thereby diminishing their group wisdom. When rewarded for their collective performance, diversity of opinions and the group wisdom increased. We conclude that the intuitive association between individual incentives and individualist opinion needs revising.

Research field(s)
Applied Sciences, Information & Communication Technologies, Artificial Intelligence & Image Processing

DOI
10.1371/journal.pone.0224725

  • NOMIS Foundation 2025
  • 2025 NOMIS Foundation