Summary of The Varieties of Instrumental Rationality
Stephen Ellis argues that the concept of instrumental rationality fixes more than one standard for prescription, evaluation and description of action. At the core of instrumental rationality is a notion commendation, as Gibbard (1990) noticed. Also at the core of instrumental rationality is the relation between one’s actions and one’s ends. Moreover, given that the same action can produce different consequences in different circumstances, instrumental rationality should also be concerned with the circumstance or situation in which the action occurs. So, Ellis develops the following general function to model instrumental rationality: R(E, S)=A. This says that the relation between an end and a situation determines an action. Ellis notes that we need to make the function more specific to capture the variety of things we want to do with it--i.e., prescribe or evaluate human behavior, and science can in turn ask whether the specific versions of the function are descriptively adequate.
Key Take Aways
This is a summary of Stephen Ellis' article.
Instrumental rationality is about taking the appropriate means to one's ends.
When evaluating or describing humans, we care about different things: what their goals were, what the situation was, what their beliefs were, and what their understanding was.
These different aspects show that instrumental rationality has different standards.
Varieties of Instrumental Rationality
The first distinction Ellis makes is in regards to what we count as the situation--the S in R(E, S)=A. In one sense, the situation might be the facts on the ground, irrespective of what the person’s beliefs are. In another sense, the situation might include the person’s beliefs. An example Ellis uses is trying to find his wife at the library and failing to do so. In the first sense, we want to know whether or not going to the library achieved his goal. In the second sense, we want to know whether, given his beliefs, he could expect that walking to the library would achieve his goal.
Another distinction Ellis makes is in regards to one’s ends--the E in R(E, S)=A. Here, one can consider basic ends or derived ends. Basic ends are things one wants for their own sake (e.g., happiness). Derived ends are things one wants for their role in achieving basic ends (e.g., money).
Uses for Instrumental Rationality
In evaluating an action, we might want to know if the action (irrespective of an agent’s beliefs) would reasonably achieve an end. Or we might want to know if the action, given an agent’s beliefs, would reasonably achieve an end. These are two different things we care about in evaluating an action and so these are different standards of instrumental rationality.
A Neglected Dimension of Instrumental Rationality
Ellis points out that, when reasoning, people compartmentalize. That is, they don’t actively tend to the totality of their beliefs and desires. They tend to a subset of them. For example, in buying groceries, humans tend to some factors (taste, price, nutrition) and not others, e.g., farm worker issues. Ellis puts this in terms of Schick’s concept of “understanding.” One is understanding the situation in a certain way--in a certain frame. So, besides one’s beliefs, one’s understanding of a situation also matters to what action is taken. Ellis thinks the concept of an understanding is descriptively useful, but is it normatively useful? Ellis thinks so, because it makes a difference in our evaluations whether the agent acted according to his or her understanding of the situation or by some other means. (Someone who makes a bad purchase based on a reasonable misunderstanding makes a different type of mistake than someone who makes a bad purchase based on an unreasonable misunderstanding).
The conclusion Ellis draws is that, with these distinctions, our normative and descriptive theorizing can become more fine grained. Moreover, we can see that there isn’t one single standard of instrumental rationality. There is a variety.
Gibbard, Allan. 1999. Wise choices, apt feelings: a theory of normative judgment.
Ellis, Stephen. 2008. "The Varieties of Instrumental Rationality". The Southern Journal
of Philosophy. 46 (2): 199-220.