Ideal Observations

1. Introduction

From the way biologists tend to talk, it seems like biology involves two very different types of phenomena simultaneously. DNA is used to produce proteins by complex molecular interactions, for example, yet in doing so it is also said to “code for” those proteins. When a neuron in the brain depolarizes, it’s said to be passing along a “message.” When a gazelle leaps high into the air, it is “signaling” to others around it. In all these cases, everyone agrees that there is a physical process unfolding according to causal principles of some kind. Yet at the same time, talk of coding and signals and messages suggests that these processes also involve information. References to information and related terms are common in biological sciences covering a range of scales, from the molecular level of genetics to ethology (the study of animal behavior) and neuroscience. Information-talk, as I will call it, is a common and widely accepted feature of the language of biologists.

What is less clear, however, is how to interpret this fact. Information-talk has a number of features that appear at first glance to be puzzling. For example, biological information seems to be intimately related to yet importantly different from the physical stuff of biological phenomena: A physical thing can be said to carry different pieces of information, and the same information can appear in different physical forms. Hence, information can be “transmitted” from one physical thing to another, potentially very different, thing. On its face, these are observations with metaphysical significance: What is this “information” biologists seem so comfortable talking about, and how does it fit into the wider scientific picture of the biological world? In particular, how does it relate to the relatively uncontroversial causal or physical aspects of biology? Is information-talk just a convenient shorthand for what are in fact garden-variety physical phenomena (Sarkar 2000; 2005)? Or does information play a deeper and more essential role that cannot be accounted for in purely physical terms (Barbieri 2016)?

Why does this matter? The idea of biological systems trading in information lies at the heart of various debates—some intrinsic to science and its methods, some with wider social significance. For some, the role of information is deeply important to what distinguishes living things from nonliving matter (Stotz 2019), perhaps even a biosignature that may help us recognize life elsewhere in the universe (Walker et al. 2018). For others, talk of information is a historical relic that confuses more than it clarifies and so should be dispensed with. Some ethologists, for instance, have argued that talk of information in animal signals is misleading because focusing on an intangible “content” to signals diverts attention from important physical aspects of signal design. Instead, they argue, we should understand animal communication as the attempt by senders to manipulate receivers (Rendall et al. 2009; Owren et al. 2010). Some even argue that information-talk should be resisted on political grounds: the idea that genes carry information, some believe, descends from outmoded and harmful ideas about the primacy of genetic factors over others in development—that certain traits are “genetically encoded” and hence immutable (Francis 2003).

What we have, then, is a question with a metaphysical flavor and that concerns the biological world and the study of it. In this chapter, I offer a way to address this question of how to interpret information-talk in biology. Importantly, my discussion goes further than an analysis of the logical or linguistic relationships between information and other concepts intrinsic to biological theory; it is not just an exercise in metaphysics of science, in the sense discussed in the introduction to this volume. Instead, it has implications for how the use of that concept drives the success of biological practices—specifically, practices of investigating and explaining how biological things solve what I will call coordination problems. Because of this, it qualifies as a work of scientific metaphysics in the sense outlined in the introduction to this volume and represented throughout other chapters. That is, it has implications for not just what biological theory is like but what the biological world is like.

I take the view (alongside other contributors to this volume) that such questions are best done with an explicit and well-defined purpose to motivate and constrain our theorizing. To that end, my particular aim here is to make sense of biological information in a way that suits what I henceforth call naturalized epistemology. The “naturalism” I refer to here is one that allows scientific developments to place constraints on other areas of our thinking. With that in mind, naturalized epistemology is the project of situating epistemic concepts—including belief, knowledge, and meaning—within a scientific picture of the world, potentially reshaping those concepts in order to make them fit. If we want to continue to understand ourselves as knowing subjects with beliefs about the world that can be true or false, a commitment to naturalism compels us to reckon seriously with science’s picture of humans as physically embodied organisms evolved to solve physical, biological, and social problems.

Many take the concept of information to play a vital role in fitting these two pictures together. On the one hand, information as a concept has its origin in mentalistic talk: it invokes epistemic ideas like instruction or evidence. On the other hand, as seen previously, information appears to be commonplace in respectable scientific discourse about biological phenomena. Because information appears to stand astride these two realms, it’s thought that we may be able to narrow the conceptual chasm between them by tracing the origins of “full-blown” mental phenomena to biological processes that are more generally informational (Dretske 1981; Millikan 2004; Skyrms 2010; Sterner 2014; Garson and Papineau 2019).

To achieve this, however, one must first lay the aforementioned metaphysical groundwork: we need to define exactly what is meant when we say that biological phenomena are trading in “information.” To do the work of naturalizing mental or epistemic notions, this account of biological information should satisfy at least two closely related criteria.

First, the account must imply that biological systems are “really” informational in a literal or substantive sense (Collier 2008). In other words, information should not be merely something we project onto what are in fact plainly physical phenomena. To use a classic example, the rings of a tree are often said to carry information about its age, but this is apparently just information for us, the observers. Instead, what we need is a sense in which something is information for the organism itself, in some sense independently of us. Without this, we lack a basis for recognizing epistemic phenomena in the natural world because “information” remains confined to the minds of the biologists.

Secondly, we need a clear sense of how information so defined relates to the causal aspects of biological phenomena. Everyone agrees that causality is vitally important to our understanding of how biology works. The key question is what talk of “information” adds to that picture—what explanatory role it can play—that ordinary causal/physical language cannot do by itself. Some have addressed this challenge by effectively reducing the latter to the former—that is, explicating information in causal terms. Holly Andersen (2017), for instance, expresses information as a measure of pattern over a causal structure. Elsewhere, recent and widely discussed work by Griffiths and colleagues has analyzed biological information simply as a measure of causal specificity—of the extent to which a cause yields fine-grained or precise control over its effect (Griffiths et al. 2015; Stotz and Griffiths 2017; Stotz 2019; see section 3.3). Yet for the particular purposes of reconceiving epistemic phenomena in biological terms, this reductive approach seems at best incomplete: What does causal specificity have to do with knowledge and related concepts? It’s proven difficult to answer that question in a way that naturalism would deem acceptable.

Taken together, these interwoven criteria establish the puzzle that an adequate characterization of information in biology should answer: What we need is a way of understanding biological information that is both (1) metaphysically integrated into the established scientific picture of biology and (2) playing a substantive explanatory role within that picture.

While there is still much debate about exactly how to solve this puzzle, there has emerged a broad (though far from universal) agreement about what the notion of biological information is broadly about; namely, that it has something to do with use. This general idea is represented in a wide variety of works in a wide variety of fields. For example, it is represented in the teleosemantic approach to representational content of Millikan (2004; 2013), Neander (1995; 2017), and others, which (very roughly) ties the content of a representation to the biological function of that representation’s consumer. In short, something means what it has the function of meaning within a system evolved to solve adaptive problems. Variations on this general theme of information are also present in Dretske (1988), Maynard Smith (2000a; 2000b), Collier (2008), Skyrms (2010), Shea (2007; 2013), Bergstrom and Rosvall (2009), Anderson and Rosenberg (2008), Seyfarth et al. (2010), Robinson and Southgate (2010), Kight et al. (2013), and Lean (2014), to name just a few. It is also strongly represented in the ecological approach to cognitive science (Gibson 1966; 1979; Chemero 2003; Baggs and Chemero 2018), which treats the environment as containing information that is perceived by the organism as affordances for achieving its own biological aims. Similar appeal to use and purpose is at work in the context of cognitive science: Bechtel (2009), for one, argues that the proper conception of “information” is the one offered by control theory, which is concerned with the design of systems that exert control over factors affecting those systems by sensing and responding to changes in those factors. This covers an extremely diverse range of works, whose similarities and differences are too many and too diverse to discuss here. The key point at present is that they broadly agree on at least one thing: namely, they treat “information” as something that is being exploited by an agent (or something like an agent) to achieve some functional purpose.

I will call this general point of agreement the use-consensus, and this consensus will be the central focus of this chapter. As is sometimes but not always acknowledged, this is inherited directly or indirectly from Peirce’s theory of signs and might be seen as an extension of Peirce’s account of meaning to the living world in general (Short 2007). One way these views differ is in how exactly they understand the idea of function that fixes the meaning of a representation or even whether the “information” in question is best thought of as meaningful at all. I use the term use-consensus as an attempt to remain as neutral as possible about these issues—to capture what is shared between these approaches and to abstract from their differences. Rather than proposing yet another variation of this view, I offer a reconstruction of the use-consensus from the perspective of practice as an operational principle that organizes certain kinds of scientific investigation.

A key feature of my account is that it is closely modeled on Woodward’s (2003; 2010) interventionist analysis of causation, which similarly aims to understand causal reasoning as a cognitive tool designed to serve various scientific aims. Ereshefsky and Reydon (2023) see Woodward’s framework as similar in spirit to their grounded functionality account (GFA) of natural kinds: in their terms, reasoning about causes serves various scientific functions in ways that are grounded in the world. With this in mind, one might think of this chapter as something like a grounded functionality account of biological information—one that is explicitly compared and contrasted with Woodward’s account of causal reasoning.

A benefit of this method of analysis, I hope to show, is that it brings to the foreground answers to the preceding conditions for a substantive account of biological information: First, it lays in sharp relief how this information is related to yet importantly distinct from the causal features of those phenomena. Specifically, it finds that causation and information need not be ontologically distinct in any strong sense; rather, they correspond to different ways of thinking about a given system for different purposes. Second, it shows when and why information can be said to be “literally” or substantively at work in biological phenomena. The answer it gives to this question embodies a form of naturalism in the spirit of Price (2011): since scientific observers are themselves living things, information “for us” is simply a special case of information “for the organism.” In a sense, then, all information is biological information.

The chapter proceeds as follows. Section 2 outlines the interventionist account of causation as developed by Woodward, highlighting key features that I will borrow and adapt in order to explicate informational reasoning in similar terms. Section 3 then develops this framework for understanding informational reasoning: First, I characterize it in general terms that include non-biological cases of informational thinking using a simple case of scientific measurement. There I introduce the concept of an ideal observation, which makes sense of the informational aspects of a system and how these differ from both its causal and statistical features. Then I show how the same framework can be used to ground a substantive view of biological information. Section 4 concludes.

2. Causal Reasoning: Control and Intervention

In this section, I outline key features of the interventionist account of causation, particularly the version defended by Woodward (2003; 2010; 2014)—henceforth ICW. As we will see, this framework is founded on explicitly pragmatic ideas: it aims to understand causality by asking what purposes the use of the concept serves in scientific practice rather than aiming to ground causality in a particular scientific or metaphysical theory. This focus on the function of causal reasoning, rather than on some intrinsic nature of causes as such, is embodied in the way it understands the commitments made by causal claims. It is this method of analysis that I aim to replicate in my analysis of the concept of information in the next section.

The first and most important feature of ICW is that it takes causality to be intimately tied to the notion of control: humans are not passive observers of our reality; we act in the world to effect changes, and our idea of causality guides reasoning about the differences our actions can make. Jenann Ismael (2017) expresses this motivation for interventionism in terms of affordances:1 “To the embedded agent who doesn’t just observe, but also intervenes in, his environment, the world is chock-full of opportunities and affordances. The terms in which he represents the world will be designed to disclose them. Causal relations are the generic form of these opportunities and affordances” (117).

The idea that causal reasoning reveals affordances for control is built into the way ICW analyzes claims about causal relationships. Consider three variables connected by edges, representing a physical system (Figure 4.1). These variables form a directed acyclic graph or DAG:

In the preceding graph, the edges connecting W and X and X and Y represent the idea that W causes X and X causes Y. (The language of familial relationships is useful here: W is a “parent” of X, and X and Y are “descendants” of W.) Interventionists in general hold that no amount of probabilistic or statistical information about these variables can express the idea that they are causally related: there is no way to express that X is a cause of Y, for instance, in terms of conditional dependencies between the variables. Instead, to say that X is a cause of Y is to say that Y would change if X were manipulated from outside the system—that changing X would change Y.

Hence, causality on this view is essentially tied to the idea of intervention. Pearl (2009), whose work informs ICW, characterizes interventions algebraically: an intervention on X “breaks” its upstream dependencies on its parents and turns it into an independent variable. In contrast, a peculiar feature of ICW in particular is that it characterizes interventions as themselves causal processes of a very particular kind, which Woodward calls ideal interventions: An ideal intervention on X with respect to Y is “a causal process that changes the value of X in an appropriately exogenous way, so that if a change in the value of Y occurs, it occurs only in virtue of the change in the value of X and not through some other causal route” (Woodward 2003, 94). Treating interventions as themselves causal processes makes it possible to explicitly represent them, as shown in Figure 4.2.

Hence, for Woodward, to claim that X is a cause of Y is to claim that an ideal intervention on X changes Y, or at least the probability distribution over Y. Since the intervention detaches X from its causal dependency on its parent W, as seen in Figure 4.2, any remaining correlation between X and Y must be due to a causal relationship between them. This view of the meaning of causal claims aims to account for how scientists actually test and refine their causal models of the world (see Woodward 2014 for more detail).

As well as distinguishing causes from non-causes, this framework allows us to differentiate between causes of some effect along a number of different dimensions (Woodward 2010). For example, one cause variable is more specific than another if it exerts more fine-grained control over the value of the effect variable; that is, if it has a greater number of values that each specify a particular value of the effect. As mentioned previously, it is this causal specificity that Stotz and Griffiths (2017) claim underwrites the notion of informational relationships in biology—an issue to which I’ll return in section 3.3.

ICW has met with several criticisms, the answers to which will be useful for understanding what follows. One class of objections centers around the view that it is overly anthropomorphic and subject-focused. We generally want to understand causality as something objective that actually governs nature independently of us and our beliefs about it, it is argued. Given that, it may seem that defining causality in terms of interventions implies that there are no causes in the natural world without agents who act on it. However, there is a way to understand how causal claims on ICW can be understood to be objective—that is, meaningful and true in a way that is, in an important sense, independent of the subject(s) making the claims. Since I will be adapting this aspect of ICW in my complementary analysis of information, I will take some time to address it.

The objectivity of causal claims in ICW comes from the fact that the interventions they posit are idealized. To idealize in scientific theorizing is to introduce assumptions that are false or unrealistic as a useful theoretical device. With that in mind, what false assumptions are made with respect to ideal interventions? Firstly, the ideal interventions central to ICW are essentially counterfactual: they posit differences in one and the same event if the value of a variable were different. This is an idealization because, by definition, it is not possible to realize and compare a set of mutually exclusive counterfactuals; instead, the best we can do is approximate this counterfactual experiment by repeating a process under controlled conditions in which only the (putative) cause is varied. Because they are counterfactual, then, ideal interventions are hypothetical (Woodward 2003, 40).

Secondly, ideal interventions are ideal in the sense that they are assumed to have an ideal degree of surgical precision: they are both causally and statistically unrelated to any other variable in the system. This is an idealizing assumption because it is always at least logically possible that there is a confounding factor that we have not controlled for or that our intervention had some other unintended effect. The best we can do, then, is to approximate these ideal interventions by controlling as much as possible. This account fits well with various aspects of scientific practice with respect to testing causal claims: For example, I’ve argued elsewhere (Lean 2020) that to maximize binding specificity in drug design is effectively to approximate an ideal intervention on the drug target.

It is these idealized features of interventions that render the associated causal claims objective in a specific but important sense. In particular, it makes sense of the idea that one thing can cause another whether or not we actually perform the kind of interventions necessary to test this claim. We can, for example, coherently talk about causal relationships in the deep past or between distant celestial bodies despite the fact that we can’t go back in time and lack the power to divert the course of planets: when we discuss these causal relationships, we are discussing hypothetical interventions on the causes, not actual ones. For example, to say that a meteorite caused the extinction of the dinosaurs means that if, say, the meteorite had been diverted to miss Earth, the mass extinction would not have occurred. This has the important semantic consequence that the truth of causal claims is not relativized to the actual abilities and actions of agents: they are about manipulability in principle, not necessarily in practice.

In addition to making causal claims independent of our abilities, this also makes them independent of our beliefs: even if our beliefs about the causal structure of the world are based on well-executed controlled experiments, those beliefs can still be wrong. This is because the experiments we actually perform to test causal claims can only be approximations of the counterfactual experiments that those causal claims are about, and so actual experimental results do not logically entail the truth or falsity of causal claims. Of course, this disconnect between the meaning of causality and the empirical observations we take as evidence of causal connections is central to Hume’s discussion about the concept. Nevertheless, causality is central to our reasoning about the world despite these worries (as Hume observed), and ICW provides a rich account of this central role as it manifests in scientific practice.

Broadly, then, the effect of idealizing the interventions posited in causal claims is that it removes from those interventions any explicit reference to agency or purposeful action. It is “heuristically useful,” Woodward says, to think of an ideal intervention as the action of an agent; however, it is possible to characterize ideal interventions without endowing them explicitly with agency; instead, we can simply stipulate causal and statistical conditions. If an action is a relationship between an agent and the system on which they act, successful control depends on both. In idealizing the role of the agent in that relationship—that is, granting all the necessary statistical and causal features of the intervener—the absence of an effect can therefore be attributed to a lack of causal power in the target variable, not to any failure to intervene properly on it. This feature, in my view, strikes a careful balance between acknowledging the essential role of agency and purpose in causal reasoning while maintaining a sense of contact with an external reality. This, I take it, is why Ereshefsky and Reydon (2023) connect ICW to their grounded functionality account of natural kinds: While causal reasoning—indeed, all reasoning—is tied to the purposes for which we as agents engage in it, in doing so we are tapping into features of the world on which we depend for those purposes to be satisfied. (See Bausman 2023 for an exploration of this “tapping into” notion in terms of adaptation.)

To summarize, ICW begins by supposing that causal reasoning is designed to serve certain purposes and then develops its account in virtue of what purposes it serves and how. First, it holds that we distinguish causal from noncausal relationships because only the former are affordances for control and that causal models are designed to reveal these affordances in a general-purpose format. Second, causal information is expressed—can only be expressed—as information about how a system would change under various interventions from outside it. Third, the interventions contained in the meaning of causal claims are idealized in various ways in order to lend objectivity to our causal models; we aim to make claims that are true or false independently of our beliefs about a system or our ability to intervene on it. These hypothetical interventions are impossible to perform; nevertheless, in forming and testing causal hypotheses, we aim where possible to approximate those ideal interventions through controlled experiment.

3. Informational Reasoning: Coordination and Observation

The previous section sketches the key features of Woodward’s interventionist framework for causality that I will adapt in developing a general account of information of the kind represented by the use-consensus. A key lesson of ICW, as I understand it, is that the concept of causality is pre-theoretic: Calling a relationship causal per se does not commit to some specific theory, from physics or somewhere else, of what grounds or underwrites that relationship. Instead, what we mean when we make such a claim is that it is a potential “lever” for bringing about changes. Of course, much of scientific inquiry involves figuring out what kind of levers they are—for example, by developing mechanistic models for how that phenomenon is realized. However, its being causal qua causal is independent of these specific details. What’s more, focusing in on these details does not reduce away the “intervention” aspect: Firstly, because intervention remains essential for distinguishing the relevant from the irrelevant properties in those models with respect to the phenomenon in question. Secondly, because abstraction is a vital aspect of explanatory generalization even when the more concrete details are available (Levy and Bechtel 2013).

In short, I take causality in ICW to be more than just a “thin concept” (Cartwright 2005)—that is, more than a placeholder for a range of richer, more informative, context-dependent “thick” concepts. Instead, I take causality in the interventionist sense to be an indispensable abstraction: it is the feature that those relations all share despite their differences—a feature that is vitally important to agents who interact purposefully with the world around them.

This idea will also be an overarching motivation for the complementary account of information that I develop here. There is already precedent in the literature for the analogous approach to “information”—that it need not have a predefined theoretical underpinning in order to be useful for empirical inquiry. This view is held by Beckett Sterner (2014), for example: following Bergstrom and Rosvall (2009), Sterner argues that it would be antithetical to biological inquiry to define “information” in a way that establishes its extension based on a priori theoretical principles about semantics. Instead, the concept of information serves simply as a diagnostic tool: it is understood in a way that guides and constrains how we gather and organize empirical cases. Given this, the only conceptual assumptions we should bring into our empirical inquiry are those that tell us how to recognize informational processes when we see them. Crucially, those processes once recognized can and should reciprocally update our theoretical notions about what biological information amounts to.

In a similar vein, Kelle Dhein (2020) details a fascinating case study illustrating how and why ethologists ascribe semantic content to behavior. His case focuses on the study of navigation by ants of the genus Cataglyphis, who are surprisingly adept at finding the shortest possible route back to their nest despite long and meandering outward journeys. A research program built around these insects has been highly successful at explaining the ants’ navigation ability: as it turns out, ants achieve this through a process of “path integration”—by storing a vector representing the distance and direction from the nest that is constantly updated during its journey. It is important to note, however, that the scientists did not begin this process with a particular theory about semantic information and how it manifests in neurophysiology. Instead, the first step is simply the identification of a surprisingly reliable match between an animal’s behavior and its circumstances—in this case, the ant’s consistent choice of the shortest journey home. Given a surprising observation of this kind, the search for an explanation is guided by the notion that the animal must possess some means of reliably choosing a successful behavior. The search for an explanation for this phenomenon is understood as the search for an adequate information channel exploited by the ant, whatever precise form that channel turns out to take.

I will return to these works in due course. For now, the key point is that, as Sterner and Dhein both illustrate, the use of “information” in these biological contexts does not depend on some precise scientific or philosophical definition of the concept. Instead, it is an operational principle that guides inquiry into a certain kind of biological phenomenon. The phenomenon in question is what I will henceforth call coordination—a concept I will elaborate later. For now, I intend this term to capture, as generally as I can muster, any kind of fortuitous or adaptive “match” between a functional entity or process and its circumstances—one that is surprising enough to invite investigation into how that match comes about. “Information” refers to whatever turns out to play a particular role in that process of coordination. A significant feature of the account I develop is that it elaborates on this idea in terms adapted from Woodward’s analysis of causal reasoning: informational reasoning pertains to coordination in the same way that causal reasoning pertains to control.

One implication of this is that biological information (in the sense shared by the use-consensus) is not fundamentally different from the everyday human uses of “information” that we tend to find less controversial; rather, they are all instances of a general type of reasoning that I characterize here. In other words, information “for us” (human observers) is simply a special case of information “for the organism.” To drive this point, I will take a detour from explicitly biological examples and develop this account of informational reasoning, and its relationship with causal reasoning, in a human context. Following that, I will return in section 3.3 to the case of information in biological practice to show how it exemplifies that same form of reasoning.

4. Conclusion

The methodology by which this chapter relates causation and information aims to embody the lessons of what is sometimes called the “practice turn” (Soler et al. 2014): that is, it aims to consider scientific reasoning as essentially practical, as something we do in order to achieve goals in a certain context. This approach is also taken by ICW: it is based on the notion that concepts like “cause” serve certain purposes and can only be fully understood by considering those purposes. In turn, then, the very origin of the notion of a cause—the reason we have such a notion in the first place—is because we are actively engaged with the world we’re trying to understand. Causality, as we understand it, is intimately connected to action: causes are things that can in principle be acted on to influence the world.

I intend my analysis of information in this chapter to be not just analogous but complementary with that view: just as causal reasoning ties to action, informational reasoning ties to perception: Information is the world’s influence on us, insofar as it guides or constrains our inferences and decisions. Just like ICW, however, the aim of objectivity in scientific inquiry leads us to consider information as something that is “out there,” in a sense, independently of those inferences and decisions. To meet this need, claims about information can be understood as claims about what a system offers to an idealized observer—one that is free of the contingencies and limitations of actual agents. This, I’ve argued, is complementary to ICW’s use of idealized interventions, which serve to express causality as an objective feature independent of actual interveners.

From a naturalist point of view, this conclusion about the complementarity of causal and informational reasoning is unsurprising: Perception and action are the two means by which agents (qua agents) interact with the world around them, and so it makes sense for each to be associated with a distinct type of reasoning engaged in by those agents. Yet those types of reasoning, while in a sense distinct, are intimately related because perception and action are ultimately inseparable: The two jointly constitute an iterative feedback loop, and hence each can only be understood in connection with the other (Dewey 1896; Hurley 2001). Nevertheless it is possible and often useful to decompose that overall process and to view it either from the causal or informational perspective; that is, to view a system either as a target of action or as a conduit for the information that guides action. The fact that biological sciences often treat their objects in at least quasi-agential terms—as doing things for reasons—underwrites what is distinctly biological about biological information.

Notes

1. 1. Ismael’s use of “affordance” is evidently quite different from its use in ecological psychology—a notion I cannot discuss at length here. It should also be noted that Ismael does not subscribe to Woodward’s particular interpretation of interventionist causation.

2. 2. However, we can allow a statistical relationship with other variables pre-observation. For example, when one variable screens off another, it may be relevant to point out that the screened-off variable offers no information to some observer over and above what has already been acquired. This may make information claims relative to some prior state of an observer, but it needn’t make them subjective.

3. 3. Of course, this evolved system means that light polarization then constrains ants’ direction of travel with high probability, which can be understood in causal and statistical terms. However, the geometric relationship holds independently of any evolution in the ant’s sensory capacities and is the reason why those sensory capacities evolved in the first place. Hence, there are two different relations at work here even if the relata are superficially similar.

4. 4. For this reason in particular, the framework I develop should not be taken to lean heavily on the use of directed graphs, which are primarily designed to represent probabilistic relationships. Other means of graphically representing this relationship may be appropriate for other types of coordination phenomena.

References

1. Andersen, H. 2017. “Patterns, Information, and Causation.” Journal of Philosophy 114, no. 11: 592–622.
2. Anderson, M. L., and G. Rosenberg. 2008. “Content and Action: The Guidance Theory of Representation.” The Journal of Mind and Behavior 29: 55–86.
3. Baggs, E., and A. Chemero. 2018. “Radical Embodiment in Two Directions.” Synthese. Berlin: Springer.
4. Barbieri, M. 2016. “What Is Information?” Philosophical Transactions of the Royal Society A 374: 20150060.
5. Bausman, W. C. 2023. “How to Infer Metaphysics from Scientific Practice as a Biologist Might.” In From Biological Practice to Scientific Metaphysics, edited by W. C. Bausman, J. K. Baxter, and O. M. Lean. Minneapolis: University of Minnesota Press.
6. Bechtel, W. 2009. “Constructing a Philosophy of Science of Cognitive Science.” Topics in Cognitive Science 1: 548–69.
7. Bergstrom, C. T., and M. Rosvall. 2009. “The Transmission Sense of Information.” Biology and Philosophy 26, no. 2: 159–76. https://doi.org/10.1007/s10539-009-9180-z.
8. Cartwright, N. 2005. “Causation: One Word, Many Things.” Philosophy of Science 71, no. 5: 805–20.
9. Chemero, Anthony. 2003. “Information for Perception and Information Processing.” Minds and Machines 13: 577–88.
10. Collier, J. 2008. “Information in Biological Systems.” In Handbook of the Philosophy of Science, Vol. 8: Philosophy of Information, edited by P. Adriaans and J. van Benthem, 763–87. Amsterdam: Elsevier Science.
11. Cropley, D. H. 1998a. “Towards Formulating a Semiotic Theory of Measurement Information—Part 1: Fundamental Concepts and Measurement Theory.” Measurement 24: 237–48.
12. Cropley, D. H. 1998b. “Towards Formulating a Semiotic Theory of Measurement Information—Part 2: Semiotics and Related Concepts.” Measurement 24: 249–62.
13. Dewey, J. 1896. “The Reflex Arc Concept in Psychology.” Psychological Review 3, no. 4: 357–70.
14. Dhein, K. 2020. “What Makes Neurophysiology Meaningful? Semantic Content in Insect Navigation Research.” Biology and Philosophy 35, no. 52.
15. Dretske, F. I. 1981. Knowledge and the Flow of Information. Cambridge, Mass.: MIT Press.
16. Dretske, F. I. 1988. Explaining Behavior: Reasons in a World of Causes. Cambridge, Mass.: MIT Press.
17. Ereshefsky, M., and T. A. C. Reydon. 2023. “The Grounded Functionality Account of Natural Kinds.” In From Biological Practice to Scientific Metaphysics, edited by W. C. Bausman, J. K. Baxter, and O. M. Lean. Minneapolis: University of Minnesota Press.
18. Finkelstein, L. 1994. “Measurement and Instrumentation Science: An Analytical Review.” Measurement 14, no. 1: 3–14.
19. Francis, R. 2003. Why Men Won’t Ask for Directions: The Seductions of Sociobiology. Princeton, N.J.: Princeton University Press.
20. Garson, J., and D. Papineau. 2019. “Teleosemantics, Selection and Novel Contents.” Biology and Philosophy 34, no. 36.
21. Gibson, J. J. 1966. The Senses Considered as Perceptual Systems. Boston: Houghton-Mifflin.
22. Gibson, J. J. 1979. The Ecological Approach to Visual Perception. Boston: Houghton-Mifflin.
23. Griffiths, P. E., et al. 2015. “Measuring Causal Specificity.” Philosophy of Science 82: 529–55.
24. Hurley, S. 2001. “Perception and Action: Alternative Views.” Synthese 129: 3–40.
25. Ismael, J. 2017. “An Empiricist’s Guide to Objective Modality.” In Metaphysics and the Philosophy of Science: New Essays, edited by M. H. Slater and Z. Yudell. Oxford: Oxford University Press.
26. Kight, C. R., et al. 2013. “Communication as Information Use: Insights from Statistical Decision Theory.” In Animal Communication Theory: Information and Influence, edited by U. E. Stegmann. Cambridge: Cambridge University Press.
27. Kosso, P. 1992. “Observation of the Past.” History and Theory 31, no. 1: 21–36.
28. Lean, O. M. 2014. “Getting the Most Out of Shannon Information.” Biology and Philosophy 29, no. 3: 395–413.
29. Lean, O. M. 2019. “Chemical Arbitrariness and the Causal Role of Molecular Adapters.” Studies in History and Philosophy of Biological and Biomedical Sciences. https://doi.org/10.1016/j.shpsc.2019.101180.
30. Lean, O. M. 2020. “Binding Specificity and Causal Selection in Drug Design.” Philosophy of Science 87, no. 1: 70–90.
31. Levy, A., and W. Bechtel. 2013. “Abstraction and the Organization of Mechanisms.” Philosophy of Science 80, no. 2: 241–61.
32. Maynard Smith, J. 2000a. “The Concept of Information in Biology.” Philosophy of Science 67, no. 2: 177–94.
33. Maynard Smith, J. 2000b. “Reply to Commentaries.” Philosophy of Science 67, no. 2: 214–18.
34. Millikan, R. G. 2004. Varieties of Meaning. Cambridge, Mass.: MIT Press.
35. Millikan, R G. 2013. “Natural Information, Intentional Signs and Animal Communication.” In Animal Communication Theory: Information and Influence, edited by U. E. Stegmann. Cambridge: Cambridge University Press.
36. Neander, K. 1995. “Misrepresenting and Malfunctioning.” Philosophical Studies 79: 109–41.
37. Neander, K. 2017. A Mark of the Mental: In Defense of Informational Teleosemantics. Cambridge, Mass.: MIT Press.
38. Owren, M. J., D. Rendall, and M. J. Ryan. 2010. “Redefining Animal Signaling: Influence versus Information in Communication.” Biology and Philosophy 25, no. 5: 755–80.
39. Pearl, J. 2009. Causality: Models, Reasoning, and Inference. 2nd ed. Cambridge: Cambridge University Press.
40. Price, H. 2011. Naturalism without Mirrors. Oxford: Oxford University Press.
41. Rendall, D., M. J. Owren, and M. J. Ryan. 2009. “What Do Animal Signals Mean?” Animal Behaviour 78, no. 2: 233–40.
42. Robinson, A., and C. Southgate. 2010. “A General Definition of Interpretation and Its Application to Origin of Life Research.” Biology and Philosophy 25, no. 2: 163–81.
43. Rossel, Samuel, and R. Wehner. 1984a. “Celestial Orientation in Bees: The Use of Spectral Cues.” Journal of Comparative Physiology A 155, no. 5: 605–13.
44. Rossel, Samuel, and R. Wehner. 1984b. “How Bees Analyse the Polarization Patterns in the Sky.” Journal of Comparative Physiology A 154, no. 5: 607–15.
45. Sarkar, S. 2000. “Information in Genetics and Developmental Biology: Comments on Maynard Smith.” Philosophy of Science 67, no. 2: 208–13.
46. Sarkar, S. 2005. “How Genes Encode Information for Phenotypic Traits.” In Molecular Models of Life: Philosophical Papers on Molecular Biology, 261–83. Cambridge, Mass: MIT Press.
47. Seyfarth, R. M., et al. 2010. “The Central Importance of Information in Studies of Animal Communication.” Animal Behaviour. Elsevier Ltd 80, no. 1: 3–8.
48. Shannon, C. E. 1948. “A Mathematical Theory of Communication.” Bell System Technical Journal 27, no. 4: 623–56.
49. Shannon, C. E. 1949. “Communication in the Presence of Noise.” Proceedings of the IRE, pp. 10–21.
50. Shea, N. 2007. “Representation in the Genome and in Other Inheritance Systems.” Biology and Philosophy 22: 313–31. https://doi.org/10.1007/s10539-006-9046-6.
51. Shea, N. 2013. “Inherited Representations Are Read in Development.” The British Journal for the Philosophy of Science 64, no. 1: 1–31. https://doi.org/10.1093/bjps/axr050.
52. Short, T. L. 2007. Peirce’s Theory of Signs. Cambridge: Cambridge University Press.
53. Skyrms, B. 2010. Signals: Evolution, Learning, and Information. Oxford: Oxford University Press.
54. Soler, L., et al. 2013. “Calibration: A Conceptual Framework Applied to Scientific Practices Which Investigate Natural Phenomena by Means of Standardized Instruments.” Journal for General Philosophy of Science 44: 263–317.
55. Soler, L., S. Zwart, M. Lynch, and V. Israel-Jost. 2014. Science After the Practice Turn in the Philosophy, History, and Social Studies of Science. New York: Routledge.
56. Sterner, B. 2014. “The Practical Value of Biological Information for Research.” Philosophy of Science 81, no. 2: 175–94.
57. Stotz, K. 2019. “Biological Information in Developmental and Evolutionary Systems.” In Evolutionary Causation: Biological and Philosophical Reflections, edited by T. Uller and K. N. Laland, 323–44. Cambridge, Mass.: MIT Press.
58. Stotz, K., and P. E. Griffiths. 2017. “Biological Information, Causality and Specificity: An Intimate Relationship.” In From Matter to Life: Information and Causality, edited by S. I. Walker, P. Davies, and G. Ellis, 366–90. Cambridge: Cambridge University Press.
59. van Fraassen, B. C. 2008. Scientific Representation. Oxford: Oxford University Press.
60. Walker, S. I., et al. 2018. “Exoplanet Biosignatures: Future Directions.” Astrobiology 18, no. 6: 779–824.
61. Wehner, Rüdiger. 1997. “The Ant’s Celestial Compass System: Spectral and Polarization Channels.” In Orientation and Communication in Arthropods, edited by Miriam Lehrer, 145–85. Basel: Birkhäuser.
62. Wehner, Rüdiger, and M. Müller. 2006. “The Significance of Direct Sunlight and Polarized Skylight in the Ant’s Celestial System of Navigation.” Proceedings of the National Academy of Sciences 103, no. 33: 12575–79.
63. Wolpert, L. 1969. “Positional Information and the Spatial Pattern of Cellular Differentiation.” Journal of Theoretical Biology 25, no. 1: 1–47.
64. Woodward, J. 2003. Making Things Happen: A Theory of Causal Explanation. Oxford: Oxford University Press.
65. Woodward, J. 2010. “Causation in Biology: Stability, Specificity, and the Choice of Levels of Explanation.” Biology and Philosophy 25, no. 3: 287–318.
66. Woodward, J. 2014. “A Functional Account of Causation; or, a Defense of the Legitimacy of Causal Thinking by Reference to the Only Standard That Matters—Usefulness (as Opposed to Metaphysics or Agreement with Intuitive Judgment).” Philosophy of Science 81, no. 5: 691–713.