Greenbriar 2, Ogregores and agency
It’s important to understand the nature of organizational agency in order to reckon with the scale and complexity of the institutions shaping public, private, and planetary futures. My thinking has evolved since writing Defining Agency in 2022. It’s time for an update.
Context
During the recent dinner conversation about tech & myth at the Greenbriar Inn sponsored by Dale Hatfield (post #1), we debated whether organizations could be agents. I’m particularly grateful to Paul Diduch, Nathan Schneider, and Niko Kovacevic for their comments. I also exchanged email afterwards individually with Dale, Niko, and Paul. This note tries to summarize a few key points. Any misrepresentations, errors, and omissions are mine.
Agency defined
There are many, often incompatible, definitions of agency. To me, agency refers to goal-directed, self-generated behavior of an entity interacting with and adapting to its environment (cf. my blog post Defining Agency).
Intentional agency, the standard view
There is an extensive philosophical literature on agency in general and shared or collective agency in particular. According to the standard conception, the relevant way of acting for an agent is to act with intention: “a being has the capacity to exercise agency just in case [that is, if and only if] it has the capacity to act intentionally.”
Minimal agency
Concepts like intentionality are a useful explanatory tool even if one remains agnostic about whether it is actually present. Dennett’s intentional stance posits that when predicting an entity’s behavior, we can choose to view it at various levels of abstraction. He suggests that one can understand human behavior at the level of the intentional stance, that is, treating a human as a rational agent, without making commitments to the reality of underlying psychology such as the reality of beliefs and desires. I see this as a defensible position on agency, too: treating organizations as agents helps us derive the “maximum information with the least cognitive effort,” to use Eleanor Rosch’s criterion (1978:28, references at the end; cited by Wikipedia:Intentional stance).
Behavior not cognition
I do not take mental states, notably self-reflective intention, as a requirement for agency. They may indeed be present, as we assume they are for human and probably animal agents. However, I believe one can explain agentic behavior in terms of non-conscious beliefs and desires, taken to mean representations of the external world and the entity’s needs, respectively. In a very simple example: a mechanical thermostat using a bimetallic strip represents “belief” about temperature as degree of bending of the strip, and its “desires” for a certain temperature as the set points of the contacts representing the warmest and coolest acceptable temperature.
Agency is a behavior—a way of acting—not an innate property. That capacity to act comes and goes; for example, an unconscious person cannot act and does not, at that time. have agency. (In contrast, being alive is a property that continues uninterrupted regardless of the state of the being concerned—until, of course, it ceases to be alive.)
I therefore take a broad view of agency that does not require representational mental states, such as the “minimal agency” of Barandiaran et al. (2009). Thermostats, bacteria, humans, and organizations all act coherently over time, across decision contexts, exhibiting feedback-based behavior that satisfies goals like homeostasis, survival, and growth. Note that, in these cases, intentionality is either absent or projected from the outside. Their agentic behavior continues even as constituent elements like components, molecules, cells, and employees change.
Put another way, I take an instrumentalist rather than realist view of agency (cf. Mark Sprevak’s entry in The Encyclopedia of the Mind). When I ascribe agency to ogregores below, it’s a way of thinking for generating predictions and doesn’t imply a commitment (as it would for a realist) to the existence of agency in ogregores.
I believe organizations evince more than the minimal agency of thermostats and bacteria, though less than the intentional agency of humans. Let us therefore now consider ogregores.
Ogregore agency
I use the term “ogregores” to name the phenomenon of organizations that exhibit agency. (See, e.g., my Ogregore analogies and Ogregore vs. organization.) Only some corporations of any given legal form—and only some organizations, more generally—exhibit agency. That is, not all organizations are ogregores, but ogregores are agents by definition.
Some corporations are so small that, pace the fiction of legal personhood, it is the owner/managers who are the agents, not the corporation. Legal status is orthogonal to corporate agenthood.
I believe some organizations exhibit the behavior of agents. I’m agnostic about whether they have intentions. I talk about organizational agency not as a metaphysical claim, but rather as a functional orientation.
Examples of corporate agency
Fatal disasters, where corporations cause large-scale death, are perhaps the most striking example of corporate agency.
The 1984 Bhopal disaster is considered the world's worst industrial accident. It was caused by many interlinked factors, including neglected safety protocols, inadequate maintenance and training, cost-cutting leading to poor safety standards, poor design of safety features, and weak regulatory oversight. Both corporate entities and individuals were held to account. In 1989, U.S.-based Union Carbide Corporation (UCC) agreed to pay US$470 million for damages caused in 1989. In 2010, seven former employees of Union Carbide India Limited (UCIL) were convicted of causing death by negligence. They were each sentenced to two years' imprisonment and fined US$2,700 (in 2023 dollars); all were released on bail shortly after the verdict.
The opening lines of Philip Pettit (2007) puts tragically fine point on the question of corporate vs. individual action (footnote references omitted):
The Herald of Free Enterprise, a ferry operating in the English Channel, sank on March 6, 1987, drowning nearly two hundred people. The official inquiry found that the company running the ferry was extremely sloppy, with poor routines of checking and management. “From top to bottom the body corporate was infected with the disease of sloppiness.” But the courts did not penalize anyone in what might seem to be an appropriate measure, failing to identify individuals in the company or on the ship itself who were seriously enough at fault. As one commentator put it, “The primary requirement of finding an individual who was liable . . . stood in the way of attaching any significance to the organizational sloppiness that had been found by the official inquiry.” [Pettit cites Eric Colvin, “Corporate Personality and Criminal Liability,” Criminal Law Forum 6 (1995): 17 as the source of the quotations.]
The two recent Boeing 737 MAX crashes are a recent example. Explanation in terms of individual people may be possible and may even provide a (potentially deceptive) sense of understanding, but it is often not sufficient. I believe it’s more productive, e.g. in policy terms, to explain these events in terms of Boeing prioritizing profit over engineering (through the tenure of several CEOs and over several decades) and regulatory capture of the FAA, than to itemize who did what, when.
On a mundane level, the recent investment spree in AI data centers by several large tech companies shows actions that don’t depend on who’s in charge. These organizations are responding in similar ways to changes in their common environment. It’s an example of the DeLanda substitution test mentioned below: does substituting people in the authority structure leave policies and routines unchanged?
Internal feedback loops
Dale pointed out that things get interesting with ogregores when internal actors start treating the collective as if it has intentions. Once people orient their actions around expectations about what “Legal is going to say” or “Finance wants”—and, I would add, expectations about the environment, like “the Market demands”—the collective begins to manifest goal-pursuing tendencies that are socially reinforced and increasingly self-fulfilling. Agency, in this view, emerges from feedback loops between internal expectations, observable individual and collective behaviors, and the environment.
How collective agency emerges
I don’t have firm beliefs about how agency arises in organizations. It may be some kind of emergence, though this is a complicated topic, especially in social ontology. Many factors will contribute, including the number of members, their interconnectedness and value alignment, organizational structure, internal and external incentives, processes and technologies used, and coordination mechanisms.
My working hypothesis is that collective, organizational agency becomes discernible through durable structures that permit sustained goal-directed behavior. These include feedback loops, institutional memory, procedures/protocols, and technologies. As Dale nicely put it, ogregoric agency arises when a collective possesses the means to act coherently over time, across decision contexts, and despite turnover in its individual members.
The collective can act in ways that wouldn’t change if people in specific roles were changed, along the lines of DeLanda (2006:37): “a large organization may be said to be the relevant actor in the explanation of an interorganizational process if a substitution of the people occupying specific roles in its authority structure leaves the organizational policies and its daily routines intact.” (I call this the DeLanda substitution test.)
As Dale pointed out, ogregores are not simply large firms or legal entities, but dynamic constellations of people, incentives, and systems that, under certain conditions, exhibit a kind of emergent agency. He noted that Heinz Pagels, in The Dreams of Reason, suggests that complexity gives rise to unintended patterns—what David Chalmers might call weak emergence—and that our frameworks for governance, ethics, and science must evolve to reflect that reality.
“Ogregore” vs. “corporation”
The “ogregore” term captures something that conventional language about “corporations” too often flattens or misdirects. Invoking Ronald Coase, Dale noted that organizations emerge when the costs of using the market exceed the costs of internal coordination. In doing so, we often create new internal dynamics—hierarchies, incentive misalignments, feedback loops, responsibility gaps—that distance decision-makers from consequences. (Decision-makers are not limited to executive officers, or even humans. They include all the people in an organization as well as automated processes.) It’s not just that ogregores cause harm; their very structure makes attribution difficult and accountability elusive.
Parts and Wholes
Niko, I think, was skeptical about ascribing agency to groups or collectives, arguing that they do not cohere into wholes in the same way individuals do. For example, he argued that a member of Congress adds up to more than the sum of their parts (brain, liver, heart, etc.) but Congress is merely the sum of its parts, i.e., it is a collection of members working under a set of laws. He suggested that "Congress voting" is shorthand for –and categorically different from—“Member voting,” which is not shorthand.
I would submit that “Member voting” is also shorthand, in Niko’s sense, for a huge raft of conscious and unconscious considerations for each member that goes into their action. I agree that it’s fair to say that say “the Member voted on a bill” means something different from “Congress voted on a bill.” However, the relevant behaviors change as we go from members to Congress as a whole. The Member votes, but Congress passes the bill—something that a member cannot do on their own, and indeed, something that many (and often most of them, as in the Senate passing the One Big Beautiful Bill) are unhappy about.
Paul has reservations, I think, about taxonomizing social entities (he thinks of them as technologies) in the same way as biological entities because they contain incoherent parts. A Linnean taxonomy certainly doesn’t work for social entities since they overlap in membership and function, but a hierarchy isn’t the only way to organize things. Further, even a single human being is incoherent, in the sense of lacking an orderly arrangement, in both physiological and mental ways. Different organs compete against each other for resources even in healthy individuals, let alone in pathologies like cancer and auto-immune disease. Our minds also consists of conflicting impulses and layers, from Plato’s representing the soul as a charioteer and two horses; to Freud’s Superego, Ego, and Id; to Daniel Kahneman’s System 1 and System 2 in his book Thinking, Fast and Slow.
Summary
This post explores the concept of organizational agency, arguing that some organizations—which I’ve dubbed ogregores—can exhibit goal-directed, adaptive behavior that constitutes a form of agency distinct from individual human agency. I take an instrumentalist approach, focusing on observable behavior rather than requiring conscious intention. I use industrial disasters and corporate investment patterns to argue that organizations can act in ways that transcend individual decision-making. The post argues that while ogregores may lack the intentional agency of humans, they exhibit more than the minimal agency of simple systems like thermostats, representing a middle ground that has important implications for understanding corporate responsibility and governance in complex institutional environments.
References
Barandiaran, X. E., Di Paolo, E., & Rohde, M. (2009). Defining Agency: Individuality, Normativity, Asymmetry, and Spatio-temporality in Action. Adaptive Behavior, 17(5), 367–386. https://doi.org/10.1177/1059712309343819
DeLanda, M. (2006). A New Philosophy of Society: Assemblage Theory and Social Complexity. Bloomsbury Publishing. https://www.bloomsbury.com/uk/new-philosophy-of-society-9780826491695/
Pettit, P. (2007). Responsibility Incorporated. Ethics, 117, 171–201.
Rosch, E. (1978). Principles of Categorization, pp. 27–48 in Rosch, E. & Lloyd, B.B. (eds), Cognition and Categorization.
No comments:
Post a Comment