So-called 'world Bayesian' versions of decision theory, such as Richard Jeffrey's, do not impose an instrumental conception of rationality because they permit agents to have preferences over their own actions
Preferences over actions(what world Bayesian theories allow agents to have)
Your direct desires about what you want to do, separate from what you want to achieve as a result.
Richard Jeffrey(the statement is about his theory)
A 20th-century philosopher who studied how people update their beliefs when they learn new information, and he developed an important theory about it called probability kinematics.
agents(referring to people in this philosophical discussion)
People, or more broadly, any thinking being capable of having beliefs and making decisions.
instrumental conception of rationality(Elster's version of rational choice theory)
A conception of rationality (also called homo economicus) according to which actions are valued and chosen not for themselves, but as more or less efficient means to a further end
rationality(Traditional conception being challenged by epistemic relativists)
A cognitive virtue and hallmark of the scientific method, intimately tied to requirements of consistency, justification, warrant, and evidence for beliefs.
Elster does not draw as sharp a distinction as he might have between the commitment to methodological individualism and the commitment to rational choice theory. Indeed, he also assumes that the latter flows directly from the former. The version of rational choice theory that Elster endorses, however, is one that is based upon a traditional instrumental (or homo economicus) conception of rationality, according to which “actions are valued and chosen not for themselves, but as more or less effici