Strategic decision-making involves cognitive challenges which are not fully understood. Until where do the currently available psychological micro-foundations of strategy research carry us, where do we need new micro-foundations? Does psychological research offer those additional micro-foundations, or do we need to pose new questions? If yes, which are those questions? During day 3, there will be room to address these questions systematically.
In the organization and strategy literature, the problem of the complexity of strategic decision-making has often been addressed. For instance, scholars have demonstrated that the complexity of business strategies can prevent entrants from successfully imitating incumbents, as the complexities of learning are too high. However, we see a tendency that in some strategy and organization papers causal explanations (like: particular decision rules in complex environments are superior to others, as they exploit the specific statistical properties of complex environments) are confused with qualitative analogies (like: strategic decision-making is like finding peaks on NK landscapes).
We may wish to develop systematic, mathematical formulations of some questions which are perhaps too vaguely defined in the strategy literature at the moment.
In particular: What characterizes the decision-environment of strategic decision-making (like the development of a new business model, or the shaping of a new industry architecture), in comparison to repeated decision situations (like the selection of an applicant form a set of applicants)? This could be pinned down by formal a specification of both sets of problems. To find a mathematical, systematic description of these decision and interaction problems would already be one great outcome of the workshop, because on this basis, one could decide what a good model of strategic decision-making is, and what not, and could also help to reduce frictions in the literature. A good decision rule may be one that accords to the 'scissors' principle suggested by Herbert Simon. This principle suggests that particular decision rules are not universal tools that can be applied everywhere, but a good decision rule needs to 'fit' to a specific decision environment (like the two blades of a scissor). Thus, before we can talk about the 'fit' of rules of strategic decision-making, we first need to pin down the properties of a strategic decision environment.
Moreover, an under-explored issue is how 'big steps' in strategic cognition are made. Gavetti (2011) proposes that big steps in strategic cognition are made by finding new mental representations of strategic decision problems. While we applaud to Gavetti's paper, and argue that the questions Gavetti asks are very important, we argue that we are lacking micro-foundations to understand the 'big steps' in strategic cognition that Gavetti talks about. Mental model theory (Johnson-Laird, 1983, 2010), often cited in the strategy literature, seems not to provide micro-foundations to explain the desired properties of strategic cognition. Mental model theory explains how individuals generate models of a situation when they are exposed to a description of a world in experiments, and how they make inferences using these models. Mental model theory does not explain differences in the possibilities that individuals see, or differences in their languages. However, mental model theory research provides experimental designs that might be further developed to study questions dear to strategy research in the future.
One question that appears to be unanswered in psychological research as well is why there are individual differences in the possibilities that different individuals see. How does 'organizational language' relate to the possibilities in an environments that individual in such organizations see?
One subtle issue is more more conceptual and fundamental. The strategy community often still draws on the distinction between 'rational' and 'behavioral' theories. However, as pointed out for instance by Johnson Laird et al. (2004), different perfectly rational agents can come to different conclusions. One reason is that is one allows competing descriptions of the world, contradictions can appear. Thus, the rationality principle alone has also theoretical limitations. Often, one cannot derive predictions about behavior from assuming that the agent is 'rational'. How agents deal with contradictions is thus an interesting question, because this elicits the beliefs of concrete agents.
While descriptive behavorial theories have little value for guiding decisions in new domains where previous rules may no longer apply, so far, normative rational theories are making unrealistic assumptions regarding the knowledge that the actors possess and the complexities of the decision situation. The question then is whether there can be any rational strategy for seeking novel solutions in unexplored directions without squandering resources by exploring too many directions in vain. In learning theory, this is known as the exploration-exploitation dilemma, and there have been attempts from information theory to address this dilemma. What can techniques from machine learning and statistics contribute here?
Furthermore, scholarly work in the field of organization and strategy suggests that beliefs of firms are often grounded in their identity construction (Prahalad and Bettis, 1986). An example are the decision patterns of the music publishers (Universal, BMG, EMI, MGM, Sony) when they faced the digital music revolution. Some patters of actions (e.g. to sue all start-up firms that attempted to sell .mp3 online, instead of embracing these efforts and buying these companies, compare Levy, The Perfect Thing, 2006) were, judged in hindsight, of clear
disadvantage to the music publishers. These decision patterns may be explained as outcomes of the identity construction of the music publishers. The publishers desperately attempted to save a business model which they probably knew was unsustainable.
Beliefs which are based on identity constructions may appear as 'biases' or delusions on individual level, but their properties are more complex. To a certain extent, biases grounded in identity construction can be self-confirming, if they help to stabilize certain structures of reality. Phrased in more radical words: We should perhaps no longer model beliefs as simply trying to represent some outside reality, but rather realize that the economic world for a participant is essentially constituted by the beliefs and expectations of the other agents. A crucial question then is when and why certain regimes of self-confirming beliefs are forced to break down, and how this process can be triggered and exploited.
The discussion how decision theory can account for the mentioned dimensions of strategic decision making will go hand in hand with the discussion of the relation of language capabilities, imagination, and unawareness (see the 11am discussion on day 2).
Please see this PDF file for the complete schedule.
January 23 - 26, 2012
Max Planck Institute for Mathematics in the Sciences
see travel instructions
Max-Planck-Institut für Mathematik in den Naturwissenschaften
London Business School
Director, Max-Planck-Institut für Mathematik in den Naturwissenschaften
University Ca' Foscari, Venice
Max-Planck-Institut für Mathematik in den Naturwissenschaften
Contact by Email