Putting dual-system theories to the test-insights from philosophy, psychology and neuroscience
Date: January 31 - February 2, 2013
Convenors: Kirsten Volz (Tübingen, GER), Elizabeth Irvine (Tübingen, GER)
How do we make decisions? Based on a range of empirical evidence it has been observed that our everyday and professional decisions can rarely be considered rational, but are often detrimentally influenced by our intuitions and emotions (Kahneman, 2003, but see also Keren and Schul, 2009). Accordingly, over the past two decades there has been an increase in theoretical frameworks suggesting the existence of two different processing systems: System 1 which is considered to be fast, intuitive and emotional and System 2 which is considered to be slow, deliberative and logical. Recently, the dual-systems approach has also been transferred to the neural level, i.e., different neural correlates were tied to an intuitive and to a deliberative system (e.g. Lieberman, 2000).
Despite the appeal of a dual-systems approach, there are a number of methodological and conceptual worries. First, there is a concern about how distinct the systems really are, what the empirical basis for this distinction may be, whether the systems are neurally plausible, and how they might interact. Relatedly, there are questions about whether there is actually an overarching dual-system theory, and if so, what role dual-system talk plays in theorizing in psychology. Further, there are questions about the possible implications of dual-system theorising on philosophical accounts of moral reasoning and rationality. Thus, by bringing together psychologists, cognitive neuroscientists and philosophers we aimed to put dual-system theories to the test.
Wim De Neys (CNRS, Université Paris Descartes) started by providing a variety of empirical evidence that during reasoning, individuals implicitly detect that their heuristic (incorrect) response conflicts with traditional normative considerations. This kind of conflict detection is incompatible with both serial and parallel models of reasoning. Instead, de Neys suggested a type of "logical intuition model" that cues both a logical and a heuristic response, where conflict can be detected without prior engagement of the deliberate system. As a commentator, Andreas Glöckner (University of Göttingen) initiated the discussion by asking how distinct heuristic and intuitive cues are, how they are integrated and how conflict might be detected. By outlining his parallel constraint satisfaction (PCS) network model, he argued that intuition may provide ongoing guidance during decision making.
Keith Frankish (University of Crete) provided an overview of criticisms of dual-systems theories, subsequent moves towards a more flexible dual-process approach. Frankish then outlined his alternative dual-level account, where type 1 processes are personal level activities (e.g. reasoning), while type 2 processes are sub-personal level activities. This included an account of inner speech and mental rehearsal that brings together many typical features of type 2 processes, and explains how they could be uniquely human. As commentators, Amadeus Magrabi (Charité Berlin) and Achim Stephan (Osnabrück University) explored this account further, in particular what the personal/sub-personal distinction can add to existing accounts.
Starting off the second day of the workshop, Jason Alexander (London School of Economics) explored the implications of dual-process theories for moral philosophy. He argued that the existence of two rational systems could undermine two main features of moral philosophy. First, what one ought (morally) to do is traditionally seen to be linked to what is rational. Yet if we do not always have control of which system of rationality directs our actions, this poses a threat to personal autonomy, which in turn threatens how we attribute moral responsibility. Comments from Liz Irvine (CIN, Tuebingen) noted similar philosophical worries about how to locate personal control, with further discussion focusing on ways out of the problems posed by Alexander.
Christine Stelzel (Charité Berlin) presented the first results of a large-scale study on the neuro-cognitive components of self-control (traditionally a feature of System 2). Stelzel argued that investigating self-control requires a diverse task battery to assess different components of self-control, and the exploration of how different components are related to each other. First results indicate that the components of emotion regulation and cognitive flexibility are associated with self-control on both the behavioral and brain level. Laura Mega (CIN, Tübingen) initiated the discussion by asking whether dual-system models can account for the trainability of self-control, and if trainability is compatible with measures of self-control as a personality trait (e.g. self-report) used in the study. Other sub-functions and measures used in the study then became the target of general discussion.
Etienne Koechlin (Laboratoire de Neurosciences Cognitives, Paris) addressed the problem that in everyday decision making we continuously have to arbitrate between staying with the current strategy, switching to another learned strategy, or exploring and possibly creating a new strategy. His findings, based on a combination of computational modeling and fMRI, suggest that in volatile environments participants use cognitive strategies that can be described as fast and computationally frugal (intuitive/System 1), whereas in stable and predictable environments they come to rely on information about category uncertainty (deliberate strategy/System 2). Thorsten Fehr (University of Bremen), while very supportive of the model, raised general questions about its ecological validity, and the methodological problems in exploring how such mental processes are realized in the brain.
Guy Kahane (Oxford Uehiro Centre for Practical Ethics) offered a detailed critique of dual-systems approaches in moral psychology (notably by Greene). He argued that claims of the existence of separate deontological and utilitarian processes for moral judgment are based on problematic test cases, and are simply inconsistent with what actually motivates people's apparently 'utilitarian' or 'deontological' judgments. Roland Deutsch's (TU Dresden) comment focused on the central points that studies based on trolley problems really are uninformative about decision making processes, and the subsequent discussion focused on the general contributions (or otherwise) of the dual-systems framework.
Our final speaker, Elisabeth Norman (University of Bergen), used empirical findings based on the concept of 'fringe consciousness' (related to metacognitive feelings), to raise problems for dual-process accounts of learning. Using an artificial grammar learning task with two separate grammars, she showed that subjects who had only implicit knowledge of the grammars were nevertheless capable of switching between the two grammars in a strategic way. This then is an example of a type 1 process (implicit learning) having distinctly type 2 properties (being available for strategic control). Her commentator Chloë FitzGerald (University Geneva) explored the concept of metacognitive feelings further, with questions in the discussion focusing on the measures used by Norman, and whether to expect a clear distinction between type 1 and type 2 processing.
The workshop therefore fulfilled our aim of putting dual-systems theories to the test. Many participants noted how productive and inclusive the interdisciplinary discussions were, much of which was owed to the particular atmosphere of the ZiF, which gradually led to some (perhaps surprising) conclusions. First was the recognition that in providing a common research framework, dual-system approaches have benefited psychology by allowing easier communication across fields, and by shaping theoretical debates. However, it also became clear that it is best to evaluate specific dual-systems/process theories individually, as they can differ widely. In this case, it makes sense to ask if there are enough similarities across specific theories to ground a 'generic' but theoretically interesting dual-system/process framework (many participants were sceptical of this). In this case, one should also be wary of drawing strong philosophical conclusions about reasoning and rationality (especially in the domain of morality) from such a generic dual-system framework.
- Kahneman, Daniel. (2003). A perspective on judgment and choice: mapping bounded rationality. The American Psychologist, 58(9), 697-720.
- Keren, G., & Schul, Y. (2009). Two Is Not Always Better Than One: A Critical Evaluation of Two-System Theories. Perspectives on Psychological Science, 4(6), 533-550.
- Lieberman, M. D. (2000). Intuition: A Social Cognitive Neuroscience Approach Logic of Intuition as Implicit Learning. Psychological Bulletin, 126(1), 109-137.
Jason McKenzie Alexander (London, GBR), Wim de Neys (Paris, FRA), Roland Deutsch (Dresden, GER), Mario Donick (Rostock, GER), Josef Ehrenmüller (Wien, AUT), Thorsten Fehr (Bremen, GER), Chloë FitzGerald (Genf, SUI), Keith Frankish (Heraklion, GRE), Andreas Glöckner (Bonn, GER), Helena Hermann (Zürich, SUI), Ninja Katja Horr (Tübingen, GER), Tobias Huber (Spiez, SUI), Guy Kahane (Oxford, GBR), Etienne Koechlin (Paris, FRA), Daniel Labbé (Simrishamn, SWE), Michael Lissack (Naples, USA), Amadeus Magrabi (Berlin, GER), Laura Mega (Tübingen, GER), John Michael (Aarhus, DEN), Santiago Arango Muñoz (Bochum, GER), Maria-Katharina Niedernhuber (Tübingen, GER), Elisabeth Norman (Bergen, NOR), Christine Stelzel (Berlin, GER), Achim Stephan (Osnabrück, GER), Thea Zander (Tübingen, GER), Frank Zenker (Lund, SWE)