AGENCY
The following terms and their respective definitions describe agency as an emergent property of a sufficiently complex information processing agent which enables that information processing agent to (a) generate a virtual model of what appears to that information processing agent to be that information processing agents’s encompassing physical environment (and the information processing agent itself as a smaller subset of that environment) and (b) to interact with that environment (particularly through a spatially finite body which acts as an interface between the information processing agent’s observing frame of reference and that information processing agent’s larger encompassing environment) such that the information processing agent is able to achieve that information processing agent’s goals.
To view hidden text inside of the preformatted text boxes below, scroll horizontally.
(Note that, in the preformatted text boxes which describe PERCEPTION, DECISION, and EXECUTION as pseudocode functions, the time increment unit of one (1) could potentially vary in temporal length. What that one (1) represents is the discrete progression of the PERCEPTION-DECISION-EXECUTION cycle in terms of one of those three steps being completed within a finite amount of time).
AGENCY: an information processing agent appearing to itself to have some degree of self-originated control over which phenomena appear and which phenomena disappear inside of that information processing agent’s frame of reference as time elapses according to that particular frame of reference.
The above definition of agency only pertains to self-aware information processing agents. It should be noted that not all information processing agents are believed to be self-aware nor even conscious despite the fact that those information processing agents, however rudimentary they are, demonstrate the ability to learn and to achieve goals which are set for them by instructions which are programmed into their hardware (either innately or as a result of them learning or otherwise being modified by factors external to their hardware).
A more rudimentary (and general) definition of agency could be “the ability of an information processing system to behave in a manner which deploys an algorithmic and structured process of learning and decision-making (instead of operating in a relatively randomized and unstructured manner) such that specific goals are achieved.”
INFORMATION_PROCESSING_AGENT: a partition of nature which is smaller than the totality of nature and which is instantiated as some relatively finite allocation of space, time, matter, and energy and which renders perceptions, thinks decisively (using a process of elimination to reduce multiple behavior options down to exactly one option), and implements behaviors which are the outcomes to such decision-making processes as a means for that information processing agent to attain its goals.
An information processing agent is essentially a computer which is capable of learning from its experiences as time elapses along that information processing agent’s encompassing space-time continuum. In particular, that information processing agent accumulates memories which enable that information processing agent to notice patterns about which types of phenomena appear to be causally related such that the information processing agent can infer which behaviors are most probabilistically likely to enable that information processing agent to achieve its goals.
ENVIRONMENT: a set of phenomena which an information processing agent renders and perceives as being external to that information processing agent’s spatially finite body (and such that the information processing agent interfaces with its (virtual and possibly physical) environment through a frame of reference which appears to be localized to “the present moment” within the context of exactly one linear and temporally forward-moving spacetime continuum which moves from “the past” to “the future” (from the vantage of that information processing agent’s sense of what “the present moment” is)).
GOAL: a set of environmental (or self-internal) conditions which an information processing agent imagines (and which are not yet being observed by that information processing agent as features of that information processing agent’s current environment or internal hardware-software state) and which compel that information processing agent to select specific thoughts and resulting behaviors which that information processing agent thinks will be sufficiently likely to cause those idealized internal or external conditions to manifest as real phenomena in that information processing agent’s world model (and hence be rendered in that information processing agent’s frame of reference) at a specific “point” in that information processing agent’s future.
An example of rudimentary goal-oriented behavior is a C++ program which sets an int type variable named X to some random integer using a specific function which returns an int type value and then, if the value of X is not equal to the “goal state” of 9, that random integer function will be called at least one more time such that the value of X will be set to the value returned by that function until the value of X is equal to the “goal state” of 9 (essentially through an iterative process of elimination).
int GOAL = 9; // The GOAL value does not determine exactly how many times the while loop in the program will be executed when the code is executed by a computer. Instead, the GOAL value constrains the computer's behavior such that the while loop stops looping if and when the GOAL value is obtained by X. int X = generate_random_integer(); while (X != GOAL) X = generate_random_integer(); // The generate_random_integer() function is called indefinitely many times until the value stored in X is equal to the value stored in GOAL.
(A more intelligent process for attaining the goal state of X in the code above is an algorithm which incorporates rudimentary learning by using a modified version of the generate_random_integer() function such that generate_random_integer() “remembers” which outputs it previously generated and which were not X by being fed those “failure” values as function inputs on successive calls to that function and storing those “failure” values in some array which is external to that function. What that approach does is incrementally reduce the degrees of freedom that function has to generate the undesired output with successive calls to that function).
PERCEPTION: an information processing agent rendering phenomena inside of that agent’s frame of reference as a result of that information processing agent which synthesizes incoming sensory input data from its physical environment through sensory hardware, memorized data from its data storage hardware, and possibly newly synthesized data about that information processing agent’s anticipated future based on what it remembers and perceives within a particular “present moment instant”.
The above definition includes an information processing agent imagining its (relatively immediate) future with enough definition to enable that information processing agent to experience a sense of causal continuity between its past and its future before actually calculating a particular course of action to implement (as a “holding space” where that information processing agent can “rest” between perception and implementing a decision-making algorithm). Imagining a relatively static and short-term future makes that information processing agent’s worldview seem relatively static (which is convenient for enabling that information processing agent to have sufficient processing time to implement its entire decision-making process preceding a resulting action).
/** * An information processing agent’s most * recent perception could be described as a * function whose input is that information * processing agent’s most recent execution * (or, if that information processing agent * recently acquired agency, that information * processing agent's most recent perception * could be described as a function of * pre-agential factors in that information processing * agent's temporally larger encompassing environment). */ perception := execution(time_i). // perception occurs at time_i + 1
DECISION: an information processing agent (a) imagining a finite set of multiple options for how that information processing agent could behave in order to achieve that information processing agent’s goals and then (b) using an algorithmic process of elimination to reduce that set of options down to exactly one option (and that one remaining option is what the information processing agent selects as the planned action to physically implement).
/** * An information processing agent’s most * recent decision could be described as a * function whose input is that information * processing agent’s most recent perception. */ decision := perception(time_i). // decision occurs at time_i + 1
EXECUTION: an information processing agent performing physical work by manipulating objects in that information processing agent’s physical environment (or within that information processing agent’s constituent physical hardware) in order to implement a behavior at time X (and immediately after that information processing agent imagined a set of multiple different behaviors which that information processing agent could select exactly one of to implement at time X in order to achieve that information processing agent’s goals).
/** * An information processing agent’s most * recent execution could be described as a * function whose input is that information * processing agent’s most recent decision. */ execution := decision(time_i). // execution occurs at time_i + 1
This web page was last updated on 10_JULY_2025. The content displayed on this web page is licensed as PUBLIC_DOMAIN intellectual property.