8. What I talk about when I talk about agency
I want to begin further exploring the topic of agency with a question:
Do you laugh at jokes that you know aren’t funny?
What I talk about when I talk about agency lies here.
Some obvious explanations for why we might do this: to be polite; to align with social convention; to be liked; to not offend. But what came first, the behavior — polite laughing — or the conditioning of the behavior?
Paul Bello and Will Bridewell write about the concept of agency in their essay There is No Agency Without Attention, laying out three types of agents.
Type 1 “agents” (those are definitely scare-quotes) have a fixed representation of their environment including uncertainty. It takes into account stimulus mappings which are used to determine hard-wired principles, such as utility maximization or cost minimization against a set of known variables. Examples cited are a home’s thermostat, a car’s cruise control mechanism, and an airplane’s autopilot system.
Type 2 agents are flexible in the face of ignorance or indifference. They can create a representation of outcomes and can decide across hypothetical representations. A thirsty deer choosing to drink from a stream when she smells a presumably dangerous bear nearby. This offers the Type 2 agent a level of control that is unattainable by Type 1.
Type 3 are the most complex. Not only do they commit to choices, but they do them in accordance with social norms — defined as what an agent ought to do. This level of ought allows the Type 3 agent to go beyond what Types 1 and 2 can. This level of control allows one to override ideas such as utility maximization and fixed principles that determine the actions taken by Types 1 and 2. A hungry man who forgot his wallet may take a sandwich from the deli or he may recall his morals and forego eating completely. He may forego utility at the cost of norms or he may disregard norms for utility. Type 3 has the capacity for an ought — what he ought to do in a specific situation given the facts and information, which includes his personal circumstances and personal desires.
But exactly how does this man choose between his hunger and his morals? If his hunger far outweighs his morals, surely he will steal the sandwich. But real life is far less so decisive. Different men have different thresholds of how much hunger he feels before he steals; some steal food without being hungry.
Aside from norms, Bello and Bridewell note that humans also have a capacity to commit themselves to the pursuit of goals. We use these larger, longer-term goals to tell us which stimuli and opportunities to ignore and which to take. Without the capacity to consider norms, hypothetical consequences, and on longer-term goal-setting, that we can focus and inhibit the conditions and information that is useful to us towards those larger goals. And that it is this capacity to focus that make us Type 3 agents with a rich capacity for agency.
Returning to my original question: why do you laugh at jokes that aren’t funny?
Is it the result of Type 1 agency — a stimulus mapping to a hard-wired principle? Type 2 agency — an action tied to hypothetical circumstances? Or is it Type 3 — in accordance to social norms and longer-term goals?
Of course, there is no correct, discrete, or good/bad answer. It may be a result of all three. Laughing at your boss’ dumb joke might be a proper stimulus mapping that might earn you a promotion in the longer-term as well as the socially normative thing to do.
But what gives us human-level agency, according to Bello and Bridewell, is the capacity to (1) set one’s own goals and (2) decide if these larger goals are worth foregoing for immediate gratification — even if your goals are immediate gratification. These are the capacities Type 1 and Type 2 agents lack. They are uniquely human; and so perhaps, it is in exercising these capacities that we are most human.