When rules are over-ruled: Virtual bargaining as a contractualist method of moral judgment.

Rules help guide our behavior-particularly in complex social contexts. But rules sometimes give us the "wrong" answer. How do we know when it is okay to break the rules? In this paper, we argue that we sometimes use contractualist (agreement-based) mechanisms to determine when a rule can be broken. Our model draws on a theory of social interactions - "virtual bargaining" - that assumes that actors engage in a simulated bargaining process when navigating the social world. We present experimental data which suggests that rule-breaking decisions are sometimes driven by virtual bargaining and show that these data cannot be explained by more traditional rule-based or outcome-based approaches.

Virtue Discounting: Observability Reduces Moral Actors' Perceived Virtue.

Performing prosociality in public presents a paradox: only by doing so can people demonstrate their virtue and also influence others through their example, yet observers may derogate actors' behavior as mere "virtue signaling." Here we investigate the role of observability of actors' behavior as one reason that people engage in such "virtue discounting." Further, we investigate observers' motivational inferences as a mechanism of this effect, using the comparison of generosity and fairness as a case study among virtues. Across 14 studies (7 preregistered, total N = 9,360), we show that public actors are perceived as less virtuous than private actors, and that this effect is stronger for generosity compared to fairness (i.e., differential virtue discounting). Exploratory factor analysis suggests that three types of motives-principled, reputation-signaling, and norm-signaling-affect virtue discounting. Using structural equation modeling, we show that observability's effect on actors' trait virtue ratings is largely explained by inferences that actors have less principled motivations. Further, we leverage experimental evidence to provide stronger causal evidence of these effects. We discuss theoretical and practical implications of our findings, as well as future directions for research on the social perception of virtue.

Emotion prediction as computation over a generative theory of mind.

From sparse descriptions of events, observers can make systematic and nuanced predictions of what emotions the people involved will experience. We propose a formal model of emotion prediction in the context of a public high-stakes social dilemma. This model uses inverse planning to infer a person's beliefs and preferences, including social preferences for equity and for maintaining a good reputation. The model then combines these inferred mental contents with the event to compute 'appraisals': whether the situation conformed to the expectations and fulfilled the preferences. We learn functions mapping computed appraisals to emotion labels, allowing the model to match human observers' quantitative predictions of 20 emotions, including joy, relief, guilt and envy. Model comparison indicates that inferred monetary preferences are not sufficient to explain observers' emotion predictions; inferred social preferences are factored into predictions for nearly every emotion. Human observers and the model both use minimal individualizing information to adjust predictions of how different people will respond to the same event. Thus, our framework integrates inverse planning, event appraisals and emotion concepts in a single computational model to reverse-engineer people's intuitive theory of emotions. This article is part of a discussion meeting issue 'Cognitive artificial intelligence'.

Does big data serve policy? Not without context. An experiment with in silico social science.

The DARPA Ground Truth project sought to evaluate social science by constructing four varied simulated social worlds with hidden causality and unleashed teams of scientists to collect data, discover their causal structure, predict their future, and prescribe policies to create desired outcomes. This large-scale, long-term experiment of in silico social science, about which the ground truth of simulated worlds was known, but not by us, reveals the limits of contemporary quantitative social science methodology. First, problem solving without a shared ontology-in which many world characteristics remain existentially uncertain-poses strong limits to quantitative analysis even when scientists share a common task, and suggests how they could become insurmountable without it. Second, data labels biased the associations our analysts made and assumptions they employed, often away from the simulated causal processes those labels signified, suggesting limits on the degree to which analytic concepts developed in one domain may port to others. Third, the current standard for computational social science publication is a demonstration of novel causes, but this limits the relevance of models to solve problems and propose policies that benefit from the simpler and less surprising answers associated with most important causes, or the combination of all causes. Fourth, most singular quantitative methods applied on their own did not help to solve most analytical challenges, and we explored a range of established and emerging methods, including probabilistic programming, deep neural networks, systems of predictive probabilistic finite state machines, and more to achieve plausible solutions. However, despite these limitations common to the current practice of computational social science, we find on the positive side that even imperfect knowledge can be sufficient to identify robust prediction if a more pluralistic approach is applied. Applying competing approaches by distinct subteams, including at one point the vast TopCoder.com global community of problem solvers, enabled discovery of many aspects of the relevant structure underlying worlds that singular methods could not. Together, these lessons suggest how different a policy-oriented computational social science would be than the computational social science we have inherited. Computational social science that serves policy would need to endure more failure, sustain more diversity, maintain more uncertainty, and allow for more complexity than current institutions support.

Too Many Cooks: Bayesian Inference for Coordinating Multi-Agent Collaboration.

Collaboration requires agents to coordinate their behavior on the fly, sometimes cooperating to solve a single task together and other times dividing it up into sub-tasks to work on in parallel. Underlying the human ability to collaborate is theory-of-mind (ToM), the ability to infer the hidden mental states that drive others to act. Here, we develop Bayesian Delegation, a decentralized multi-agent learning mechanism with these abilities. Bayesian Delegation enables agents to rapidly infer the hidden intentions of others by inverse planning. We test Bayesian Delegation in a suite of multi-agent Markov decision processes inspired by cooking problems. On these tasks, agents with Bayesian Delegation coordinate both their high-level plans (e.g., what sub-task they should work on) and their low-level actions (e.g., avoiding getting in each other's way). When matched with partners that act using the same algorithm, Bayesian Delegation outperforms alternatives. Bayesian Delegation is also a capable ad hoc collaborator and successfully coordinates with other agent types even in the absence of prior experience. Finally, in a behavioral experiment, we show that Bayesian Delegation makes inferences similar to human observers about the intent of others. Together, these results argue for the centrality of ToM for successful decentralized multi-agent collaboration.

The logic of universalization guides moral judgment.

To explain why an action is wrong, we sometimes say, "What if everybody did that?" In other words, even if a single person's behavior is harmless, that behavior may be wrong if it would be harmful once universalized. We formalize the process of universalization in a computational model, test its quantitative predictions in studies of human moral judgment, and distinguish it from alternative models. We show that adults spontaneously make moral judgments consistent with the logic of universalization, and report comparable patterns of judgment in children. We conclude that, alongside other well-characterized mechanisms of moral judgment, such as outcome-based and rule-based thinking, the logic of universalizing holds an important place in our moral minds.

Nonlinearities between inhibition and T-type calcium channel activity bidirectionally regulate thalamic oscillations.

Absence seizures result from 3 to 5 Hz generalized thalamocortical oscillations that depend on highly regulated inhibitory neurotransmission in the thalamus. Efficient reuptake of the inhibitory neurotransmitter GABA is essential, and reuptake failure worsens human seizures. Here, we show that blocking GABA transporters (GATs) in acute rat brain slices containing key parts of the thalamocortical seizure network modulates epileptiform activity. As expected, we found that blocking either GAT1 or GAT3 prolonged oscillations. However, blocking both GATs unexpectedly suppressed oscillations. Integrating experimental observations into single-neuron and network-level computational models shows how a non-linear dependence of T-type calcium channel gating on GABAB receptor activity regulates network oscillations. Receptor activity that is either too brief or too protracted fails to sufficiently open T-type channels necessary for sustaining oscillations. Only within a narrow range does prolonging GABAB receptor activity promote channel opening and intensify oscillations. These results have implications for therapeutics that modulate inhibition kinetics.

What We Owe to Family: The Impact of Special Obligations on Moral Judgment.

Although people often recognize the moral value of impartial behavior (i.e., not favoring specific individuals), it is unclear when, if ever, people recognize the moral value of partiality. The current studies investigated whether information about special obligations to specific individuals, particularly kin, is integrated into moral judgments. In Studies 1 and 2, agents who helped a stranger were judged as more morally good and trustworthy than those who helped kin, but agents who helped a stranger, instead of kin were judged as less morally good and trustworthy than those who did the opposite. In Studies 3 and 4, agents who simply neglected a stranger were judged as less morally bad and untrustworthy than those who neglected kin. Study 4 also demonstrated that the violation (vs. fulfillment) of perceived obligations underlaid all judgment patterns. Study 5 demonstrated boundary conditions: When occupying roles requiring impartiality, agents who helped a stranger instead of kin were judged as more morally good and trustworthy than agents who did the opposite. These findings illuminate the importance of obligations in structuring moral judgment.

Drivers are blamed more than their automated cars when both make mistakes.

When an automated car harms someone, who is blamed by those who hear about it? Here we asked human participants to consider hypothetical cases in which a pedestrian was killed by a car operated under shared control of a primary and a secondary driver and to indicate how blame should be allocated. We find that when only one driver makes an error, that driver is blamed more regardless of whether that driver is a machine or a human. However, when both drivers make errors in cases of human-machine shared-control vehicles, the blame attributed to the machine is reduced. This finding portends a public under-reaction to the malfunctioning artificial intelligence components of automated cars and therefore has a direct policy implication: allowing the de facto standards for shared-control vehicles to be established in courts by the jury system could fail to properly regulate the safety of those vehicles; instead, a top-down scheme (through federal laws) may be called for.

People Make the Same Bayesian Judgment They Criticize in Others.

When two individuals from different social groups exhibit identical behavior, egalitarian codes of conduct call for equal judgments of both individuals. However, this moral imperative is at odds with the statistical imperative to consider priors based on group membership. Insofar as these priors differ, Bayesian rationality calls for unequal judgments of both individuals. We show that participants criticized the morality and intellect of someone else who made a Bayesian judgment, shared less money with this person, and incurred financial costs to punish this person. However, participants made unequal judgments as a Bayesian statistician would, thereby rendering the same judgment that they found repugnant when offered by someone else. This inconsistency, which can be reconciled by differences in which base rate is attended to, suggests that participants use group membership in a way that reflects the savvy of a Bayesian and the disrepute of someone they consider to be a bigot.

Lucky or clever? From expectations to responsibility judgments.

How do people hold others responsible for the consequences of their actions? We propose a computational model that attributes responsibility as a function of what the observed action reveals about the person, and the causal role that the person's action played in bringing about the outcome. The model first infers what type of person someone is from having observed their action. It then compares a prior expectation of how a person would behave with a posterior expectation after having observed the person's action. The model predicts that a person is blamed for negative outcomes to the extent that the posterior expectation is lower than the prior, and credited for positive outcomes if the posterior is greater than the prior. We model the causal role of a person's action by using a counterfactual model that considers how close the action was to having been pivotal for the outcome. The model captures participants' responsibility judgments to a high degree of quantitative accuracy across three experiments that cover a range of different situations. It also solves an existing puzzle in the literature on the relationship between action expectations and responsibility judgments. Whether an unexpected action yields more or less credit depends on whether the action was diagnostic for good or bad future performance.

Learning a commonsense moral theory.

We introduce a computational framework for understanding the structure and dynamics of moral learning, with a focus on how people learn to trade off the interests and welfare of different individuals in their social groups and the larger society. We posit a minimal set of cognitive capacities that together can solve this learning problem: (1) an abstract and recursive utility calculus to quantitatively represent welfare trade-offs; (2) hierarchical Bayesian inference to understand the actions and judgments of others; and (3) meta-values for learning by value alignment both externally to the values of others and internally to make moral theories consistent with one's own attachments and feelings. Our model explains how children can build from sparse noisy observations of how a small set of individuals make moral decisions to a broad moral competence, able to support an infinite range of judgments and decisions that generalizes even to people they have never met and situations they have not been in or observed. It also provides insight into the causes and dynamics of moral change across time, including cases when moral change can be rapidly progressive, changing values significantly in just a few generations, and cases when it is likely to move more slowly.

Statistically inaccurate and morally unfair judgements via base rate intrusion.

From a statistical standpoint, judgements about an individual are more accurate if base rates about the individual's social group are taken into account 1-4 . But from a moral standpoint, using these base rates is considered unfair and can even be illegal 5-9 . Thus, the imperative to be statistically accurate is directly at odds with the imperative to be morally fair. This conflict was resolved by creating tasks in which Bayesian rationality and moral fairness were aligned, thereby allowing social judgements to be both accurate and fair. Despite this alignment, we show that social judgements were inaccurate and unfair. Instead of appropriately setting aside social group differences, participants erroneously relied on them when making judgements about specific individuals. This bias-which we call base rate intrusion-was robust, generalized across various social groups (gender, race, nationality and age), and differed from analogous non-social judgements. Results also demonstrate how social judgements can be corrected to achieve both statistical accuracy and moral fairness. Overall, these data (total N = 5,138) highlight the pernicious effects of social base rates: under conditions that closely approximate those of everyday life 10-12 , these base rates can undermine the rationality and fairness of human judgements.

Multiple micronutrient supplementation reduces anemia and anxiety in rural China's elementary school children.

Despite growing wealth and a strengthening government commitment to improve livelihoods and welfare, many students across rural China have inadequate access to micronutrient-rich diets. Poor diets can lead to nutritional problems, such as iron-deficiency anemia, that can adversely affect health, attention, learning, and mental health. The overall goal of this paper is to assess the impact of multiple micronutrient supplementation (MMS) on anemia and anxiety among students in poor areas of rural China. To achieve this goal, we conducted a randomized controlled trial in 54 randomly chosen elementary schools in 8 of the poorest counties in Shaanxi Province in Northwest China. Study participants were 2730 fourth-grade students, mostly aged 10-12 y. Schools were randomly assigned to 1 of 2 groups: a control group that received no intervention and an intervention group that received a daily MMS with 5 mg of iron (ferrous sulfate) for 5 mo. Our primary outcome measures were hemoglobin (Hb) concentrations (assessed by HemoCue 201+ technology), anemia prevalence (defined as Hb) concentrations ≤120 g/L), and anxiety (using a written mental health test). The results showed that 42.4% of students were anemic at baseline. The Hb concentration was 121.7 ± 10.7 g/L in the treatment group and 123.4 ± 11.4 g/L in the control group. MMS increased Hb concentrations by 1.7 g/L ± 0.15 and reduced anemia rates by 7.0 percentage points (P < 0.05). Anxiety was reduced by 0.30 SDs (P < 0.01). MMS reduced both anemia and anxiety. Our results should encourage further research on the linkages between nutrition and mental health in a development context.

Rescuing the Corticostriatal Synaptic Disconnection in the R6/2 Mouse Model of Huntington's Disease: Exercise, Adenosine Receptors and Ampakines.

In the R6/2 mouse model of Huntington's disease (HD) we examined the effects of a number of behavioral and pharmacological manipulations aimed at rescuing the progressive loss of synaptic communication between cerebral cortex and striatum. Two cohorts of transgenic mice with ~110 and 210 CAG repeats were utilized. Exercise prevented the reduction in striatal medium-sized spiny neuron membrane capacitance but did not reestablish synaptic communication. Activation of adenosine A2A type receptors renormalized postsynaptic activity to some extent. Finally, the ampakine Cx614, which has been shown to prevent α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate (AMPA) receptor desensitization, slow deactivation, and facilitate glutamate release, induced significant increases in synaptic activity, albeit the effect was somewhat reduced in fully symptomatic, compared to control mice. With some limitations, each of these strategies can be used to delay and partially rescue phenotypic progression of HD in this model.

Anemia in rural China's elementary schools: prevalence and correlates in Shaanxi province's poor counties.

Despite growing wealth in China, a significant share of children across rural China still have no access to iron-rich foods, vitamins, and other micronutrients. Such poor diets may result in high incidences of nutritional problems, including anemia. The objective of the study was to increase understanding of the extent of anemia, and identify structural correlates of anemia in poor Shaanxi province's primary schools. The article shows that the overall anemia rate is 21.5 percent when using a blood hemoglobin cutoff of 115 g/L (39 percent with a cutoff of 120 g/L). We find that those students that are boarding at school and eat lunch away from home are more likely to be anemic. Children with anemia are found to have lower height for age (HAZ) scores. If this part of Shaanxi province is representative of all poor counties in China, these findings mean millions of children in poor rural China may be anemic.

A gain in GABAA receptor synaptic strength in thalamus reduces oscillatory activity and absence seizures.

Neural inhibition within the thalamus is integral in shaping thalamocortical oscillatory activity. Fast, synaptic inhibition is primarily mediated by activation of heteropentameric GABA(A) receptor complexes. Here, we examined the synaptic physiology and network properties of mice lacking GABA(A) receptor alpha3, a subunit that in thalamus is uniquely expressed by inhibitory neurons of the reticular nucleus (nRT). Deletion of this subunit produced a powerful compensatory gain in inhibitory postsynaptic response in nRT neurons. Although, other forms of inhibitory and excitatory synaptic transmission in the circuit were unchanged, evoked thalamic oscillations were strongly dampened in alpha3 knockout mice. Furthermore, pharmacologically induced thalamocortical absence seizures displayed a reduction in length and power in alpha3 knockout mice. These studies highlight the role of GABAergic inhibitory strength within nRT in the maintenance of thalamic oscillations, and demonstrate that inhibitory intra-nRT synapses are a critical control point for regulating higher order thalamocortical network activity.

Synergistic roles of GABAA receptors and SK channels in regulating thalamocortical oscillations.

Rhythmic oscillations throughout the cortex are observed during physiological and pathological states of the brain. The thalamus generates sleep spindle oscillations and spike-wave discharges characteristic of absence epilepsy. Much has been learned regarding the mechanisms underlying these oscillations from in vitro brain slice preparations. One widely used model to understand the epileptiform oscillations underlying absence epilepsy involves application of bicuculline methiodide (BMI) to brain slices containing the thalamus. BMI is a well-known GABAA receptor blocker that has previously been discovered to also block small-conductance, calcium-activated potassium (SK) channels. Here we report that the robust epileptiform oscillations observed during BMI application rely synergistically on both GABAA receptor and SK channel antagonism. Neither application of picrotoxin, a selective GABAA receptor antagonist, nor application of apamin, a selective SK channel antagonist, alone yielded the highly synchronized, long-lasting oscillations comparable to those observed during BMI application. However, partial blockade of SK channels by subnanomolar concentrations of apamin combined with picrotoxin sufficiently replicated BMI oscillations. We found that, at the cellular level, apamin enhanced the intrinsic excitability of reticular nucleus (RT) neurons but had no effect on relay neurons. This work suggests that regulation of RT excitability by SK channels can influence the excitability of thalamocortical networks and may illuminate possible pharmacological treatments for absence epilepsy. Finally, our results suggest that changes in the intrinsic properties of individual neurons and changes at the circuit level can robustly modulate these oscillations.

Differential electrophysiological properties of dopamine D1 and D2 receptor-containing striatal medium-sized spiny neurons.

The electrophysiological properties of distinct subpopulations of striatal medium-sized spiny neurons (MSSNs) were compared using enhanced green fluorescent protein as a reporter gene for identification of neurons expressing dopamine D1 and D2 receptor subtypes in mice. Whole-cell patch-clamp recordings in slices revealed that passive membrane properties were similar in D1 and D2 cells. All MSSNs displayed hyperpolarized resting membrane potentials but the threshold for firing action potentials was lower in D2 than in D1 neurons. In voltage clamp, the frequency of spontaneous excitatory postsynaptic currents was higher in D2 than in D1 cells and large-amplitude inward currents (> 100 pA) were observed only in D2 cells. After tetrodotoxin this difference was reduced, suggesting that sodium conductances contribute to the increased frequencies in D2 cells. After pharmacological blockade of GABA(A) receptors, a subset of D2 cells also displayed large spontaneous membrane depolarizations and complex responses to stimulation of the corticostriatal pathway. To further characterize ionotropic glutamate receptor function, alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionate (AMPA) was applied onto dissociated MSSNs. Application of AMPA alone or in the presence of cyclothiazide (an AMPA receptor desensitization blocker) evoked larger currents in D1 than in D2 cells. Together, these data demonstrate significant differences in electrophysiological properties of subpopulations of MSSNs defined by selective expression of D1 and D2 receptors. D2 cells display increased excitability and reflect ongoing cortical activity more faithfully than D1 cells, an effect that is independent of postsynaptic AMPA receptors and probably results from stronger synaptic coupling. This could help to explain the increased vulnerability of D2 MSSNs in neurodegenerative disorders.