Revisiting Hayek's Political Economy by Boettke Peter J.;Storr Virgil Henry;Boettke Peter J.;

Author:Boettke, Peter J.;Storr, Virgil Henry;Boettke, Peter J.;
Language: eng
Format: epub
Publisher: Emerald Publishing Limited
Published: 2016-11-28T00:00:00+00:00

Flavors of Individualism: Atoms, Robots and Sociopaths versus Actors, AIs and Agents

While modern economics is methodologically individualist, Hayek locates the origin of the term “individualism” in the work of people opposed to the individualist ethos (Hayek, 1946), the early French socialists. Methodological individualism is operationalized in economics today by having each person in the economy (or a representative person) autonomously extremize a scalar-valued utility function. Indeed, this is the workhorse of the profession. For every new economic environment the trick is to come up with a set of constraints and a utility function that when maximized in the context of the constraints yields comparative statics that comport with a researcher’s mental or empirical model of how the world works. Indeed, some researchers even believe their job is the “programming robot imitations of people” (Lucas to Klamer, 1984).

In practice this model is well known to be a poor description of humans from a behavioral perspective, with the list of cognitive biases now numbering in the hundreds (Camerer, 1995; Gigerenzer & Selten, 2001). What is more, the idea of an individual acting without regard for other agents, or at least without direct interactions with any of them but only via indirect interactions through constraints, is at least a peculiar model of humans, if not as my former Brookings colleague Henry Aaron has written, “a passable description of a sociopath” (Aaron, 1994).

If people behaved without regard for others the atomistic interpretation of methodological individualism would be valid. It would be as if people were robots.3 Yes there are economic interactions that do not require much more than effectively anonymous interactions, as Buchanan has described certain market transactions (Bowles, 1998). But other kinds of economic activity require repeated and informationally intense interactions, not ones well represented by autonomous optimization of a scalar welfare function, rather by cognitive models (Anderson, Matessa, & Lebiere, 1997; Sun, 2006).

One reason why game theory represents an advance over conventional microeconomics is because it explicitly models direct interactions between players. Each player has goals, strategies, and payoffs, both expected and realized. Austrian economists today are not major contributors to game theory proper, perhaps because they find the highly analytical focus of the field very narrow, the “solution concepts” impoverished, or the static character of play very limiting, all things that were apparent to early mathematical economists like Herbert Simon (cf. his review (Simon, 1958) of the Luce & Raiffa, 1957 text).

In artificial intelligence (AI) it is common to build agents that take account of the state, beliefs, actions, or desires of others in determining their behavior (Russell & Norvig, 2002). Indeed, there is a wide range of software agents in active use today, and they can be classified according to how much “intelligence” they have. AIs with goals and the ability to take actions toward those goals look more like the kinds of agents we typically use in agent-based computing. Indeed, the origin of multi-agent systems (MAS) computer science is largely an outgrowth of giving goals to distributed AIs (DAIs) (see Fig.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.