Group Decisions as Optimal Integration: Two Challenges
Ophelia Deroy
Collective decisions are often compared to cases of multisensory integration, where combining multiple sources of evidence leads to decisions which are systematically better than the ones that would be derived from a single source of evidence. Anecdotal and scientific evidence, however, should make us doubt that groups are always ‘wiser’ than their members. This course will cover two main issues: First, admitting that the model of multisensory integration is relevant to model collective decisions, we will show whether we can indeed predict that collective decisions will not be optimal; second, we will discuss whether the comparison is appropriate, and examine some of the key components that it leaves out, including the fact that accuracy might not be the main function of group decisions. The course will highlight new connections between philosophical literature and work in cognitive neuroscience and experimental psychology.
Reason-based Choice
Franz Dietrich
Intuitively, we decide and act based on reasons. The reasons for attending this summer institute might include the programme on offer and the cultural and natural surroundings. By contrast, standard choice-theoretic models explain behaviour not based on reasons, but based on preferences (or constructs underlying them, such as utilities and probabilities). The relationship between reasons and preferences is not obvious. For instance, it is unclear whether preference-based choice is in conflict or rather compatible with reason-based choice, and whether reasons are absent from or rather implicitly present in choice theory. This series of lectures seeks to bring light into such issues. In a first step, I will discuss different notions or accounts of “reasons” controversially discussed in philosophy, such as normative versus motivating reasons, externalist versus internalist reasons, and practical versus theoretical reasons. This analysis will lead to the conclusion that reasons (in the more relevant senses) are absent from standard choice theory. How could we take reasons on board of our models? In a second step, I will present a “reason-based choice theory”, based on my recent work with Christian List. Within this framework, I will discuss two important forms of context-dependence relevant to behavioural economics: context-related reasons (i.e., reasons about the context) versus context-variant reasons (i.e., reasons that change with the context). In a third step, I will turn to normative rather than motivating reasons, and give a reason-based model of many notorious distinctions drawn in moral philosophy, including that between consequentialism and deontology, that between universalism and relativism, that between monism and pluralism, that between atomism and holism, and that between teleological and non-teleological ethics.
The Social Epistemology of Argumentation
Catarina Dutilh-Novaes
Argumentation (the practice of giving and asking for reasons to support claims) is a key component of scientific inquiry, legal procedures, and political life. But in many instances, argumentation does not achieve its presumed goal of fostering consensus and the circulation of reliable information. To better understand these phenomena, I adopt a social epistemological approach to argumentation, which views argumentation as a form of social exchange that can be successful to various degrees. In this course, I present and develop a novel account of argumentation as the exchange of epistemic resources such as knowledge, evidence, justification, and critical objections. Beside a detailed presentation of the general framework, which combines insights from social epistemology, argumentation theory, and social exchange theory, I discuss two specific case studies: argumentation in public, political discourse, and argumentation in science. I also investigate argumentation in contexts of epistemic diversity, and occurrences of epistemic injustice in argumentative processes.
Formal Issues in Value Aggregation
Kenny Easwaran
Utilitarian ethics, democratic politics, and many other settings, require the aggregation of individual values to create some set of group values. While the role of the group aggregate is practical in the democratic case and theoretical in the utilitarian case, and there are many other important differences between them as well, there are some formal difficulties that these questions have in common. Arrow’s famous impossibility theorem shows that any association between individual rank-orderings of a set of a choices and a collective rank-ordering of those same choices will have at least one of four bad features. Utilitarians traditionally avoid this problem by assuming that value is represented not by a rank-ordering but by a numerical score, and aggregate by adding those scores, and using probabilistic expectations when there is uncertainty of an outcome. This course will discuss formal features that arise when individual values are more specific than mere rank-ordering but less specific than precise numerical scores, as well as in cases where the group consists of subsequent generations (which may be infinite). Even if problems don’t arise in practice, considering these more general formal settings can help shed light on *why* addition and expectation are the right ways to aggregate when they do apply.
Metacognitive Myopia – a Major Impediment of Rational Behavior
Klaus Fiedler & Florian Kutzner
What I have come to call “meta-cognitive myopia” (MM), using a term once suggested by Robyn Dawes, is the phenomenon that people are pretty accurate in utilizing even large amounts of stimulus information. However, they are uncritical and naïve regarding the history and validity of the given information samples. This uncritical reliance on the information given is most conspicuous when the task context makes it crystal-clear that the stimulus data should not be trusted. I locate MM within a broader framework of meta-cognition research and illustrate the phenomenon with examples from various research paradigms. MM offers an alternative account of many biases in judgment and decision making, which have been traditionally explained in terms of capacity constraints, limited reasoning ability, motivational forces, or severely biased environmental input.
The explanatory power of the MM construct, and its theoretical potential to predict new findings, is demonstrated with reference to five paradigms: inability to discard irrelevant information; utilization of selectively sampled information; conditional inference biases; sample-size neglect; and myopia for the impact of aggregation levels. The final discussion is concerned with the learning origins of MM and the question of why evolution did not equip homo sapiens with more effective meta-cognitive tools. An analysis of the costs and benefits will reveal that MM may serve important adaptive functions, and that eliminating MM may have maladaptive effects. Nevertheless, in the context of many real decision problems, the costs and irrational consequences of MM cannot be denied. The final discussion therefore focuses on possible ways to avoid and alleviate MM and its irrational consequences.
Adaptive Behavior in an Uncertain Social World
Nadine Fleischhut & Christin Schulze
Every day throughout our lives, we make choices in the face of uncertain outcomes: which career path to pursue, which apartment to rent, how much effort to put in at work, and so forth. Many of these judgements and decisions are not made in isolation, but are informed by or depend on the behavior of others. In this course, we ask how the social environment either helps or hinders people in making judgments and decisions in an uncertain world. While others can be a source of uncertainty, the social environment can also provide powerful tools for reducing uncertainty. For example, pooling incomplete and uncertain information in group-contexts can boost the accuracy of judgments and decisions beyond the capacity of each lone individual. Additionally, sampling experiences made by people in our social networks can provide insight into the social structure of the world at large. When the social environment is governed by shared social norms, these norms can both reduce social uncertainty and benefit individuals and social groups in general. Yet at the same time, the social environment can also be hostile when people compete against each other for the same goals. The course will introduce examples from different lines of research to show how the social environment either helps or hinders adaptive behavior in an uncertain world.
The Neural Basis of Social Decision-Making
Keise Izuma
In this course, we will address social aspects of rationality from the perspectives of social neuroscience and neuroeconomics. Historically, neuroscientists investigated the neural basis of decision-making based on materialistic rewards in non-social settings. However, it is well known in psychology and economics that our attitudes and behavior are influenced by a variety of social factors (e.g., intangible social rewards, concerns for others, etc.). In the past two decades, utilizing theories and empirical findings from behavioral sciences combined with cognitive neuroscience methods (e.g., fMRI), social neuroscientists and neuroeconomists have been investigating the neural basis of such social decision-making. The course will 1) cover methods and data analysis techniques used in the field, 2) discuss key findings on neural mechanisms underlying social phenomena such as altruism, cooperation, fairness, reputation and cognitive dissonance, and 3) discuss how cognitive neuroscience methods can be used to address not only questions about how the brain works (e.g., what does brain region X do), but also questions about the mind (e.g., psychological questions).
Reasoning in Context: Law, Emotions, and Interactions
L. Estefania Gazzo Castañeda & Markus Knauff
The course starts with an overview about the main theories of human reasoning. We show how different cognitive theories explain human reasoning competencies and difficulties. Some topics will be how people reason deductively, how they deal with uncertain information and probabilities, how they revise beliefs and defeat valid information in the light of new evidence, and how they account for prior knowledge, values, and beliefs in reasoning. The second session will apply the different reasoning theories to the area of legal reasoning. In such tasks, people with or without legal knowledge need to come to conclusions that are legally justified. We show that such inferences are affects by many psychological factors such as moral values and the emotional effects of a crime. The third session will be concerned with the interaction of emotion and reasoning. We will show that emotions can sometimes help and sometimes hinder cognitive reasoning. The fourth section will focus on the social and interactive aspects of human reasoning. We will explain how groups of people aggregate and balance multiple individual opinions, share information with each other, exchange arguments and counter-arguments, and come to conclusions that everyone in the group endorses. At the end of the course, we will have learned that reasoning does not take place in a vacuum, but in a rich and complex psychological and external context.
Rational Decision Making and Moral Behavior
Martin Kocher
We introduce the approach that behavioral and experimental economics has taken to understand human behavior, in particular when it comes to the deviations from the assumption related to the homo oeconomicus. The course will focus on other-regarding preferences, social norms and normative norms as drivers of behavior in economically relevant settings. We will also discuss several aspects in the context of dis(honest) and unethical behavior, whenever there is a trade-off between following a normative norm and earning more. The course puts an emphasis on (experimental) evidence, recent research papers, the discussion of open questions that we believe worth exploring in future research—thus hoping to spark your interest and creativity related to this topic.
The Wisdom of Crowds
Aidan Lyon
This course will cover the different forms of the wisdom of crowds phenomenon –– from voting, to judgement aggregation, to deliberation groups, to prediction markets. We’ll study the phenomenon from the perspectives of psychology, philosophy, mathematics, computer simulations, social policy, and business decision making. A key focus of the course will be on how you can take advantage of, and enhance, the wisdom of crowds.
Social Norms
Ryan Muldoon
Much of our behavior depends on what we think other people do and what we think they want us to do. Social norms represent an important class of behaviors that depend on this interdependence of social belief and behavior. In this course, we will examine a theoretical framework for understanding norms, as well as consider a variety of empirical examples. Finally we will consider epistemological issues in measuring and intervening on norms.
Modeling Science: Formal Approaches to Social Epistemology
Cailin O’Connor
In assessing how best to create knowledge, it is important not to ignore social aspects of scientific communities. Scientists share knowledge, collaborate, conform to group beliefs, and attend to the social identities and beliefs of their peers in deciding what evidence to trust and who to interact with. All these aspects of interaction influence the way ideas and practices spread in scientific communities. While there is a long tradition of using formal tools to understand individual knowledge and rationality, more recently philosophers of science have turned to formal methods to gain insight into the social aspects of epistemology. Employing agent-based models in particular, philosophers have shown how individual and group rationality can pull apart, investigated optimal communication networks for scientific groups, tested the effects of conformity on science, looked at how polarized beliefs might emerge in science, and gained control over what sorts of funding, hiring, and publishing structures might optimize scientific output. This course will introduce central models and results in this sub-discipline of philosophy.
Theory of Voting
Don Saari
Voting is a critical, fundamental aspect of society that, at any instant, most surely is being used somewhere in the world. Voting is how group decisions are made, leaders selected, the winner of Oscar’s best movie or Grammy’s song of the year chosen, a group’s choice of pizza for lunch determined, and even the name of a new family pet selected. Outcomes can matter, so why can they be so bad? Should the voters be blamed? Maybe, but as developed in these lectures, surprisingly often the true villain is the voting rule. In the 2016 US Presidential election, for instance, Hillary Clinton received nearly 3 million more votes than Donald Trump—and lost! Why?
The theory of voting, which is developed in these lectures, will explain these problems in an intuitive manner where the participants wiill be able to create all sorts of new paradoxes. Even more: Voting is a aggregation rule, which means that lessons learned about voting help to identify what can go wrong with other aggregation methods. Examples that I expect to discuss include statistical methods ranging from hypothesis testing to paired comparison methods that are often used in psychology, sociology, economics, and engineering. Other expected topics include how lessons from voting theory provide insights about unexpected properties of Adam Smith’s “Invisible Hand” story, issues from medical ethics, pitfalls of the commonly used reductionist method, and even that compelling astronomical mystery of “dark matter.”
Lectures will be adjusted to reflect the backgrounds and interests of the participants. No prior knowledge off any of these topics is required.
Logical Modelling of Social Epistemic Processes
Sonja Smets & Fernando R. Velazquez Quesada
In this course we use modal logic to reason about a number of interesting but puzzling social epistemic scenario’s in which agents exchange information, revise their beliefs and update their knowledge. We focus on intriguing epistemic puzzles such as the muddy children puzzle and Moore’s paradox as well as on phenomena studied in social and behavioral science such as informational cascades, pluralistic ignorance, the bystander effect etc. For instance, in an informational cascade we observe how individual agents in a sequence decide to follow the public signals/decisions of their predecessors while simply ignoring their own private evidence. Such agents are ‘following the crowd’, whether or not the crowd is right or wrong. In this context we are not looking at examples of just mindless imitation or mass hysteria, no, these agents make a rational choice. In this case we apply logic to analyse whether rational agents who use their higher-order reasoning powers can ultimately stop a cascade from happening. Logic helps us to model and reason about diffusion of knowledge and belief in a network and can help us analyse the properties of different network creation processes. The logical toolbox that is needed in this course is based on modal logic, in particular epistemic and doxastic logics are used to represent the knowledge and beliefs of agents. Later in the course we extend these logics with ingredients to represent epistemic social networks. We go over examples to get familiar with these logical techniques and on the philosophical side we refer to the latest work in social epistemology.
Neural Mechanisms of Intersubjectivity
Kai Vogeley
The presentation will cover theoretical and empirical issues of interaction and communication in social encounters from an interdisciplinary perspective covering social psychology, psychopathology, and social cognitive neuroscience. Interaction and communication will be presented on the basis of two fundamental modes of information processing including reflexive, explicit, inferential processes as well as prereflexive, implicit, nonverbal communication. Making use of virtual character technology and functional neuroimaging these processes can also be systematically studied on a neural level. These studies revealed that two different functional systems appear to constitute two different routes of processing underlying our social cognitive capacities in everyday social encounters, namely the so-called “mirror neuron system” (MNS) and the “social neural network” (SNN, also theory of mind network or mentalizing network). The functional roles of both systems appear to be complementary: The MNS serves comparatively “early” stages of social information processing that are more related to spatial or bodily signals, whereas the SNN serves comparatively “late” stages of social information processing that are more related to the “evaluation” of emotional and psychological states of others. Most recently, we focused on the study of “social gaze” and the experience of being engaged in an ongoing interaction with others. Truly interacting with others additionally recruits the reward system possibly reflecting the hedonic experience and primary motivation to interact with others. Conceptually, we will explore self-consciousness and intersubjectivity within the theoretical framework of analytical philosophy of mind. Methodologically, we will study the different psychological processes and neural mechanisms addressing the complex phenomena of self-consciousness and intersubjectivity (including mentalizing, perspective taking, agency, and nonverbal communication. Finally, we will discuss the explanatory potential of both concepts for different psychopathological conditions (e.g. schizophrenia, autism).
Network Epistemology
Kevin Zollmann
Groups communicate in many different ways, and how they communicate influences how the groups learn and perform. Many tragedies of group rationality are not the result of any individual failing, but rather a failure to communicate in appropriate ways. This course will look at how groups might communicate, how they do better and worse, and how incentives to communicate might influence them. To understand these issues we will use mathematical tools from network theory, game theory, and statistics. But, this course will not presume any prior knowledge of these fields; it will be self-contained.
