When we engage in the process of division of labor, there are typically multiple alternatives, but insufficient knowledge to choose among them. Under such conditions, we propose that not all alternatives are equally likely to be pursued. In particular, when we engage in the process of division of labor for novel and non-repetitive production, we argue that we display a tendency to perceive and select objectbased task partitions over activity-based partitions. We experimentally investigate how the salience of objects over activities manifests itself in individuals and groups engaged in division of labor for the assembly of strongly or weaklydecomposable products. We draw implications for organization design as well as the impact of technological change on organizations. The limited prior attention to the process by which a division of labor emerges may simply have been the consequence of pervasive functionalist assumptions. For instance, the influential position of Ludwig von Mises was that if a particular division of labor has efficiency advantages (such as gains from trade), its emergence could be taken for granted (von Mises 1949). This may even have been justified in an industrial context dominated by highly repetitive activity on a shop floor, such as that celebrated in Smith (1776), where opportunities for learning through repetition could have helped uncover the efficient division of labor, and opportunities for amortizing that learning through repeated production may have led 2 In particular, the process of achieving task division may be uniquely important in human organizations. Division of labor also occurs in biological systems, but the key questions there pertain to why some systems evolve towards differentiated allocation of tasks given a task division and an initial undifferentiated allocation of tasks (Rueffler Hermisson and Wagner 2012).
Page 2 of 49 Management Science3 to a strong emphasis on particular kinds of division of labor (e.g. Taylor 1911; Wren 1979). However, the degree of repeatability of production that organizations confront today varies significantly (e.g. it is relatively lower in project-based work than on the shop floor). Further, for every existing pattern of division of labor, there must have been a point in time at which the process by which it first emerged unfolded. Put simply, organizations often confront novel goals, and when they do, a process of division of labor must necessarily arise. If so, it is valuable to understand what social and psychological factors beyond the technological properties of the task influence the process through which an equilibrium -a division of labor -emerges (Schelling 1960;Skyrms 2004).In this study, we investigate this process of division of labor under conditions of novelty and nonrepetitiveness with a focus on its social and psychological micro-foundations. It has been noted that we tend to focus more on 'objects' than 'activities' when making sense of and categorizing our environment (e.g. Gentner 1982;Rosch et al. 1976). We prop...