A recent research trend following in the footsteps of Brunswik's probabilistic functionalism has been to entertain and evaluate whether human judgment and decision strategies may, in some contexts, make even fewer demands on cognition than the linear-additive (compensatory) rules presumed by the standard lens model (Dhami & Ayton, 2001; Dhami & Harries, 2001; Gigerenzer & Goldstein, 1996; Gigerenzer et al., 1999). This trend is readily apparent in the “ecological rationality” paradigm of Gigerenzer and colleagues, which portrays cognition as an “adaptive toolbox” of “fast and frugal” heuristics (for more detail, see especially chapters 10, 12, and 18 and the introduction to part VI). This paradigm shares with Brunswik the view that cognitive strategies should be evaluated according to their adaptive mesh with both environmental demands (e.g., time and accuracy constraints) and resources (e.g., available information sources and their statistical texture).
At present, however, it is premature to estimate what impact the ecological rationality approach will have in the area of human–technology interaction. Looking at the temporal aspects of the situation, there are ample reasons to believe that its impact will be significant. Technological work contexts often require working under high levels of time pressure, in which quick yet approximate solutions are likely to earn more favor than slow yet precise ones (e.g., chapter 18). To date, however, those who have adopted the ecological rationality perspective have chiefly focused on informing cognitive theory, rather than on studying and supporting behavior in the types of dynamic, interactive, technological ecologies that are the focus of this book.
As such, the first chapter in this section represents one of the first attempts to view technology interaction from the perspective of ecological rationality. In Chapter 10, Rothrock and Kirlik revisit the dynamic judgment task studied in chapter 3 by Bisantz and colleagues but from the perspective of heuristic-or rule-based (noncompensatory) rather than linear-additive (compensatory) modeling. Rothrock and Kirlik note that the ecological rationality approach currently lacks techniques for inferring rule-based heuristics directly from behavioral data (the purpose served by linear regression in compensatory lens modeling and policy capturing). To address this need, they present a study in which they developed a technique using a genetic algorithm to infer candidate noncompensatory descriptions of the logical (if/and/or/not/then) rules that best described the behavior of participants in their experiment. Notably, their findings shed additional light on the relatively high use of unmodeled knowledge