Robot swarm democracy: the importance of informed individuals against zealots

In this paper we study a generalized case of best-of-n model, which considers three kind of agents: zealots, individuals who remain stubborn and do not change their opinion; informed agents, individuals that can change their opinion, are able to assess the quality of the different options; and uninformed agents, individuals that can change their opinion but are not able to assess the quality of the different opinions. We study the consensus in different regimes: we vary the quality of the options, the percentage of zealots and the percentage of informed versus uninformed agents. We also consider two decision mechanisms: the voter and majority rule. We study this problem using numerical simulations and mathematical models, and we validate our findings on physical kilobot experiments. We find that (1) if the number of zealots for the lowest quality option is not too high, the decision-making process is driven toward the highest quality option; (2) this effect can be improved increasing the number of informed agents that can counteract the effect of adverse zealots; (3) when the two options have very similar qualities, in order to keep high consensus to the best quality it is necessary to have higher proportions of informed agents.

Attention alters certainty but not appearance

Does paying attention to a stimulus change its appearance or merely influence the decision mechanisms involved in reporting it? Recently we proposed an uncertainty stealing hypothesis in which subjects, when uncertain about a perceptual comparison between a cued and uncued stimulus, tend to disproportionately choose the cued stimulus. The result is a psychometric function that mimics the results that would be measured if attention actually changed the appearance of the cued stimulus. In the present study, we measure uncertainty explicitly. In three separate experiments, subjects judged the relative appearance of two Gabor patches that differed in contrast. In the first two experiments, subjects performed a comparative judgment, reporting which stimulus had the higher contrast. In the third experiment, subjects performed an equality judgment, reporting whether the two stimuli had the same or different contrast. In the first comparative judgment experiment and in the equality judgment experiment, one of the two stimuli was pre-cued by an exogenous cue. In the second comparative judgment experiment, a decision bias was explicitly introduced: one stimulus was followed by a post-cue and the subjects were instructed, when uncertain, to choose the cued target. In all three experiments, subjects also indicated whether or not they were certain about each response. The results reveal that in the pre-cue comparative judgment, attention shifted the subjects’ uncertainty and made subjects more likely to report that the cued stimulus had higher contrast. In the post-cue biased comparative judgment, subjects also were more likely to report that the cued stimulus had higher contrast, but without a shift in uncertainty. In the equality judgment, attention did not affect the contrast judgment, and the subjects’ uncertainty remained aligned with their decision. We conclude that attention does not alter appearance but rather manipulates subjects’ uncertainty and decision mechanisms.

Pitfalls of Machine Learning based Personnel Selection - Fairness, Transparency and Data Quality

In recent years, machine learning (ML) modeling (often referred to as artificial intelligence) has become increasingly popular for personnel selection purposes. Numerous organizations use ML-based procedures for screening large candidate pools, while some companies try to automate the hiring process as far as possible. Since ML models can handle large sets of predictor variables and are therefore able to incorporate many different data sources (often more than common procedures can consider), they promise a higher predictive accuracy and objectivity in selecting the best candidate than traditional personal selection processes. However, there are some pitfalls and challenges that have to be taken into account when using ML for a sensitive issue as personnel selection. In this paper, we address these major challenges - namely the definition of a valid criterion, transparency regarding collected data and decision mechanisms, algorithmic fairness, changing data conditions as well as adequate performance evaluation - and discuss some recommendations for implementing fair, transparent, and accurate ML-based selection algorithms.

Pitfalls of Machine Learning-Based Personnel Selection

In recent years, machine learning (ML) modeling (often referred to as artificial intelligence) has become increasingly popular for personnel selection purposes. Numerous organizations use ML-based procedures for screening large candidate pools, while some companies try to automate the hiring process as far as possible. Since ML models can handle large sets of predictor variables and are therefore able to incorporate many different data sources (often more than common procedures can consider), they promise a higher predictive accuracy and objectivity in selecting the best candidate than traditional personal selection processes. However, there are some pitfalls and challenges that have to be taken into account when using ML for a sensitive issue as personnel selection. In this paper, we address these major challenges – namely the definition of a valid criterion, transparency regarding collected data and decision mechanisms, algorithmic fairness, changing data conditions, and adequate performance evaluation – and discuss some recommendations for implementing fair, transparent, and accurate ML-based selection algorithms.

Asymmetrical Property of the Subproportionality of Weighting Function in Prospect Theory: Is It Real and How Can It Be Achieved?

An asymmetrical property of the probability weighting function, namely, subproportionality, was derived from observations. Subproportionality can provide a reasonable explanation for accommodating the Allais paradox and, therefore, deserves replication for its high impact. The present study aimed to explore the mechanism of subproportionality by comparing the two completely opposite decision mechanisms: prospect theory and equate-to-differentiate theory. Results revealed that the underlying mechanism supports the prediction of equate-to-differentiate theory but not prospect theory in the diagnostic stimuli condition. Knowledge regarding which intra-dimensional difference between Options A and B is greater, not knowledge regarding which option’s overall prospect value is greater, indeed predicts option preference. Our findings may deepen current understanding on the mechanisms behind the simple risky choice with a single-non-zero outcome. Additionally, these findings will hopefully encourage subsequent researchers to take a fresh look at the Allais paradox.

Explainable Brain Disease Classification and Relevance-Guided Deep Learning

Deep neural networks are increasingly used for neurological disease classification by MRI, but the networks' decisions are not easily interpretable by humans. Heat mapping by deep Taylor decomposition revealed that (potentially misleading) image features even outside of the brain tissue are crucial for the classifier's decision. We propose a regularization technique to train convolutional neural network (CNN) classifiers utilizing relevance-guided heat maps calculated online during training. The method was applied using T1-weighted MR images from 128 subjects with Alzheimer's disease (mean age=71.9+-8.5 years) and 290 control subjects (mean age=71.3+-6.4 years). The developed relevance-guided framework achieves higher classification accuracies than conventional CNNs but more importantly, it relies on less but more relevant and physiological plausible voxels within brain tissue. Additionally, preprocessing effects from skull stripping and registration are mitigated, rendering this practically useful in deep learning neuroimaging studies. Understanding the decision mechanisms underlying CNNs, these results challenge the notion that unprocessed T1-weighted brain MR images in standard CNNs yield higher classification accuracy in Alzheimer's disease than solely atrophy.

Neuro-computational foundations of moral preferences

Moral preferences pervade many aspects of our lives, dictating how we ought to behave, whom we can marry and even what we eat. Despite their relevance, one fundamental question remains unanswered: where do individual moral preferences come from? It is often thought that all types of preferences reflect properties of domain-general neural decision mechanisms that employ a common ‘neural currency’ to value choice options in many different contexts. This view, however, appears at odds with the observation that many humans consider it intuitively wrong to employ the same scale to compare moral value (e.g. of a human life) with material value (e.g. of money). In this paper, we directly test if moral subjective values are represented by similar neural processes as financial subjective values. In a study combining functional magnetic resonance imaging with a novel behavioral paradigm, we identify neural representations of the subjective values of human lives or financial payoffs by means of structurally identical computational models. Correlating isomorphic model variables from both domains with brain activity reveals specific patterns of neural activity that selectively represent values in the moral (right temporo-parietal junction) or financial (ventral-medial prefrontal cortex) domain. Intriguingly, our findings show that human lives and money are valued in (at least partially) distinct neural currencies, supporting theoretical proposals that human moral behavior is guided by processes that are distinct from those underlying behavior driven by personal material benefit.

The Virtuous Cycle of Agreement

Collective choice mechanisms are used by groups to reach decisions in the presence of diverging preferences. But can the employed mechanism affect the degree of post-decision actual agreement (i.e. preference homogeneity) within a group? And if so, which are the features of the choice mechanisms that matter? Since it is difficult to address these questions in natural settings, we employ a theory-driven experiment where, after the group collectively decides on an issue, individual preferences can be properly elicited. We find that decision mechanisms that promote consensual behaviour generate substantially higher levels of post-decision actual agreement compared to outcome-wise identical procedures that incentivize subjects to exaggerate their differences.

Detection of District Heating Pipe Network Leakage Fault Using UCB Arm Selection Method

District heating networks make up an important public energy service, in which leakage is the main problem affecting the safety of pipeline network operation. This paper proposes a Leakage Fault Detection (LFD) method based on the Linear Upper Confidence Bound (LinUCB) which is used for arm selection in the Contextual Bandit (CB) algorithm. With data collected from end-users’ pressure and flow information in the simulation model, the LinUCB method is adopted to locate the leakage faults. Firstly, we use a hydraulic simulation model to simulate all failure conditions that can occur in the network, and these change rate vectors of observed data form a dataset. Secondly, the LinUCB method is used to train an agent for the arm selection, and the outcome of arm selection is the leaking pipe label. Thirdly, the experiment results show that this method can detect the leaking pipe accurately and effectively. Furthermore, it allows operators to evaluate the system performance, supports troubleshooting of decision mechanisms, and provides guidance in the arrangement of maintenance.