Risk

This note is about the concept and the examples of risk. Most definitions of risk assume a human evaluator who decides which outcomes count as harm. The note records where that assumption works, then extends risk to nonhuman beings, who face, assess and produce risks of their own.

Cf.

• Uncertainty
• Redundancy
• Relational Accounting
• Ignorance
• Disaster
• Extinction
• Resilience
• Cognition
• Agency
• Anthropocentrism

Risk and Uncertainty

Risk and uncertainty describe different kinds of not-knowing.

Risk refers to situations where analysts know the main outcomes and can assign probabilities with enough confidence to support calculation.1^, 2

Uncertainty includes cases where analysts cannot specify probabilities clearly, cannot complete the causal model, must work across multiple relevant futures, or cannot fully define the set of possible outcomes in advance.3^, 1^, 4

A useful way to state the relation is: risk is a special case of uncertainty.1^, 4 Not every uncertain situation becomes a risk problem, because many situations do not have stable probabilities or a settled model.

Working distinction:

• Risk means uncertain outcomes with usable odds.1^, 2
• Uncertainty means that analysts cannot fully settle the odds, model, relevant futures, or problem framing.5^, 4
• Deep uncertainty means that even well-informed analysts cannot justify a single probabilistic representation of the future.3
• Ignorance and indeterminacy describe deeper limits where important surprises, open causal chains, or unrecognized conditions remain in play.4^, 2
• Relational uncertainty highlights that not-knowing can arise from differences between knowers, institutions, and worlds, as well as from missing data.6
• Cultural risk framing highlights that what counts as risk is shaped by social selection, institutions, and values.7^, 8
• Lived uncertainty highlights action within unstable conditions where responsiveness may matter more than prediction.9

Risk in Its Anthropocentric Guise

Calculative risk is a human practice with a clear domain of use. The modules below record how it centres human evaluators, where it still works, and how it can produce the harms it sets out to manage.

Defined by Human Value

• Textbook definition. A standard environmental-psychology definition describes risk as a situation, event or activity that "may lead to uncertain adverse outcomes affecting something that humans value". The same text adds that "characterizing an outcome as adverse involves a human evaluation".10
• Risk society. Sociology ties risk to modernity, defining it as "a systematic way of dealing with hazards and insecurities induced and introduced by modernization itself". The thesis acknowledges that modern hazards threaten "people, animals and plants".11 A posthumanist critique points out that it still refers risk entirely to humans: "it's us human beings, and only us who experience risk".12
• Risk talk as a sign of entanglement. The same critique reads talk of risk as evidence that people now imagine material flows leaving the factory, travelling through the environment and returning to affect them. Its example follows Asian eels moved from Japan to the United States: the eels grew large enough to climb out of their tanks, their owners threw them into ponds, and the eels then took food from the bass.12 Each move answered the one before, so the hazard to the bass came from human and eel actions together.
• Cultural selection. Societies select which dangers to attend to, so what counts as a risk reflects institutions and values as well as hazards.7

Ecosystems and Species as Bearers of Risk

• Climate risk. The IPCC defines risk as "the potential for adverse consequences for human or ecological systems". Its list of what may be harmed places "ecosystems" alongside lives, livelihoods, assets, infrastructure and services.13 The Sixth Assessment Report groups key risks into eight categories, one of which covers terrestrial and ocean ecosystems.14 Ecosystems therefore count as bearers of risk, although the definition treats them as one item among the goods of human societies.
• Extinction risk. The IUCN Red List converts the prospect of extinction into graded categories. A Critically Endangered taxon faces "an extremely high risk of extinction in the wild", and one quantitative criterion sets this at a probability of extinction of at least 50% within ten years or three generations, whichever is longer.15 Here calculative risk works on behalf of nonhuman beings: the species bears the risk, while human assessors set the thresholds.

Existential Risk

Centre for the Study of Existential Risk

• Definition. Existential risk refers to outcomes that "would either annihilate Earth-originating intelligent life or permanently and drastically curtail its potential".16 The bearer of risk is humankind or its possible successors.
• Ecosystems as threat. Ecosystem collapse enters this literature as a danger to humanity, as the subtitle "nature strikes back" announces.17
• Existence as an achievement. One critique observes that the field "appears to take existence as a given". It neglects that existence is "something that needs to be established and maintained".18 For living beings, maintaining existence is the ordinary condition (see Risk Relative to the Agent below).
• Long-term risks. Examples:
  • AI.
  • Locked-in injustice.

Limits of Use

• Where calculation works. Calculative risk works where outcomes are specified, probabilities can be estimated, and a named evaluator decides what counts as harm (see Risk and Uncertainty above). Insurance, engineering safety and Red List assessment meet these conditions.
• Where it misleads. Problems start when this practice claims to define risk as such. It then implies that beings unable to calculate odds cannot face risk. It also counts harm only once a human evaluator registers it.
• Where it produces risk. Flood-control levees in the Mississippi River Delta deprived bald cypress swamps of their seasonal freshwater. Channels cut for timber extraction and shipping let in saltwater, which killed further stands. Because bald cypress resist storm damage better than other vegetation, their loss left the delta more exposed to hurricanes. A 2008 simulation that removed the Mississippi River Gulf Outlet and restored the cypress forests its construction destroyed reduced Katrina flooding by 80%.19 The case shows that risk management for human settlement alone increased the flood risk to that settlement.

Nonhuman Beings Face and Assess Risk

Behavioural ecology has studied risk assessment in animals for decades. Its evidence shows that assessing risk requires neither language nor numerical probability.

• Trade-offs under predation. A foundational review finds that animals "possess the ability to (i) assess their risk of being preyed upon and (ii) incorporate this information into their decision making". These decisions trade predation risk against the benefits of feeding, escaping, socialising and, for some fishes, breathing air.20

• Landscapes of fear. Prey perceive risk as spatially uneven. The landscape of fear refers to "the spatial variation in prey perception of predation risk", which may exist as a mental map or arise in real time as an animal moves through heterogeneous risk.21

  

  ^{Framework for the landscape of fear, from the physical landscape through risk and perception to response (Fig. 1 in the source).21}

• Humans as super predators. A landscape-scale playback of human voices led large carnivores to avoid the sound and move more cautiously. Medium-sized carnivores became more elusive and foraged less. Small mammals evidently benefited, increasing their use of habitat as well as their foraging. Many wildlife impacts attributed to human activity "may be explained by fear of humans".22 Wild animals, in other words, assess human presence as a risk.

• Learned risk in a filter feeder. Blue mussels (Mytilus edulis) that had met trematode parasites before reduced their filtration when the parasites' infective stages were present, while parasite-naive mussels did not. The reduced filtration lowered infection at the cost of feeding.23 The mussels therefore traded one risk against another on the basis of experience.

• Humans as prey. An account of surviving a crocodile attack argues that death in a crocodile's jaws threatens "the dualistic visions and divisions which justify rational human mastery of the planet".24 On this reading, part of what human risk discourse protects is a picture of humans standing outside food chains.

Risk Distributed through Relations

• Entangled vulnerability. In India, the collapse of vulture populations left cattle carcasses to slower scavengers such as feral dogs, whose numbers appear to have risen sharply. Uncleared carcasses raise the threat of anthrax, while more street dogs mean more rabies, which kills mostly people from poor socioeconomic groups as well as millions of dogs.25 Vulnerability here "emerges, at least in part, as a feature of the specific ways in which we are entangled in our multispecies world". The people bound most tightly into relations with vultures and carcasses face the greatest exposure.25
• Risk already realised. Risk assessment usually looks forward from a present that has not yet suffered the loss in question. For Anishinaabek, settler colonial campaigns have "already depleted, degraded, or irreversibly damaged the ecosystems, plants, and animals" known to their ancestors. They therefore "already inhabit what our ancestors would have understood as a dystopian future".26 For many Indigenous peoples, and for many species, the losses that climate risk assessments project into the future are already history.

Risk Relative to the Agent

If nonhuman beings face risk, a definition needs to locate risk in each agent's own conditions of persistence. Two accounts from theoretical biology make this possible.

• Viability. Adaptivity is the capacity of a living system to regulate its states relative to the boundary of its viability. The system distinguishes tendencies that approach the boundary from those that recede from it, then acts on that distinction. Adaptivity thereby establishes "an inner norm, a distinction between good, neutral, and bad ways of realizing autopoiesis". It is also a graded property.27 Risk, on this account, is the prospect of approaching the viability boundary, judged by the agent's own norms.

• Scale. The cognitive light cone describes the spatial and temporal extent over which an agent can take measurements and exert influence. A tick has a small cone with little memory or prediction. A dog's cone reaches further in time, while human minds operate over much larger spans.28 Risk inherits this scale: an agent can assess only risks that fall within its cone, though risks beyond it can still harm it.

  

  ^{Cognitive light cones (Fig. 2 in the source). (A) The spatial and temporal reach of an agent's concerns. (B) Cones of different size for a tick, a dog, a human, or a possible alien or AI. (C) Nested cones of cells, organisms and colonies.28}

• Existence is maintained. Every living being must continually regulate itself to stay viable, so the possibility of ceasing to exist accompanies every organism.27 This reading extends the critique of existential-risk research above, which observes that existence needs to be established and maintained.18

• Working definition (this note's proposal). Risk is a situated relation in which uncertain conditions could interfere with an agent's efforts to establish, maintain or restore the states, processes and relations it depends on. An agent assesses risk when it differentiates such possibilities and changes its activity in response.

• What the definition allows. It requires no language, reflective judgement or numerical probability. It makes risk relative to each agent's capacities and conditions of persistence. It also covers risks that the agent never perceives.

• Limit. Risk assessment is a capacity of identifiable agents. Ecological communities can resist, recover or reorganise after disturbance without forming such an agent; see Is a forest an agent? and Resilience. Evidence of assessment comes from flexible, compensatory action, of the kind documented for animals above.

Three Formal Accounts

Three traditions make the working definition precise. Each treats the agent as the bearer of risk, yet each places risk somewhere different: in regulated variables, in predicted outcomes or in signs.

Cybernetic Risk

• Definition. Risk is the possibility that disturbances will drive the variables an agent regulates outside the range it can maintain or recover.

• Regulation. Cybernetics describes regulation in terms that apply to machines as well as organisms, and its examples range from the air-conditioner to the phototaxic plant. A regulator $F$ acts between disturbances $D$ and essential variables $E$, keeping $E$ within a permissible set $\eta$. An essential function of a regulator is "that it shall block the transmission of variety from disturbance to essential variable". Natural selection favours organisms with such regulators: "the better F is as a regulator, the larger the organism's chance of survival".29

• Requisite variety. Measured logarithmically, the variety of outcomes $V_O$ cannot fall below the variety of disturbances $V_D$ minus the variety of the regulator's responses $V_R$:

  $$V_O \geq V_D - V_R$$

  In other words, "only variety in R can force down the variety due to D".29 Risk therefore depends on the match between the disturbances an agent meets and the responses available to it. A disturbance that one agent absorbs can overwhelm another.

• Models. Under very broad conditions, "every good regulator of a system must be a model of that system".30 An agent that regulates well therefore models, in some form, the system it regulates.

• Eel example. Seaward-migrating European eels (Anguilla anguilla) in the forebay of a hydropower intake mostly followed modelled streamlines, then switched behaviour on meeting accelerating flow. The response depended on "both the novelty and strength of the transition stimuli". Most eels rejected rapid acceleration by swimming back upstream.31 In cybernetic terms, flow acceleration works as a signal of disturbance ahead of the intake, and the switch keeps the eel away from it.

  

  ^{Locations where seaward-migrating European eels switched behaviour, over modelled water acceleration, under (a) unrestricted flow and (b) constricted flow (Fig. 4 in the source).31}

• Strength. The concepts are substrate-independent, so they apply to organisms, machines and collectives alike.

• Limitation. A feedback mechanism "may instantiate a purpose, but this purpose is external to it, the mechanism by itself does not generate it".27 Cybernetic risk therefore needs an account of where the regulated range comes from. It can also mislead when agents transform their goals as well as defending them.

Active-Inference Risk

• Definition. Risk is the expected divergence between the outcomes an agent predicts under a course of action and the outcomes it prefers.

• Formal term. The expected free energy $G$ of a policy $\pi$ splits into expected ambiguity and risk:32

  $$G(\pi) = \underbrace{\mathbb{E}_{Q(s_\tau \mid \pi)}\big[H[P(o_\tau \mid s_\tau)]\big]}_{\text{expected ambiguity}} + \underbrace{D_{KL}\big[Q(o_\tau \mid \pi) \,\|\, P(o_\tau \mid C)\big]}_{\text{risk}}$$
  • $\pi$: a policy, that is, a sequence of actions.
  • $o_\tau$, $s_\tau$: outcomes and hidden states at a future time $\tau$.
  • $Q(s_\tau \mid \pi)$, $Q(o_\tau \mid \pi)$: the states and outcomes the agent predicts if it follows $\pi$.
  • $P(o_\tau \mid C)$: prior preferences over outcomes, set by parameters $C$.
  • $P(o_\tau \mid s_\tau)$: how hidden states generate outcomes. Its entropy $H$ measures ambiguity.
  • $D_{KL}$: the Kullback–Leibler divergence, which measures how far one probability distribution departs from another.

• Interpretation. Risk "quantifies the expected cost of a policy as a divergence from preferred outcomes".33 The same quantity $G$ balances seeking new information against seeking preferred observations, so caution and curiosity follow from one principle.32

• Scope. Living systems can be described as Markov blankets of Markov blankets, "reaching all the way down to cellular organelles and DNA". Only "adaptive active inference" lets an organism modulate its sensorimotor coupling to its environment.34

• Strength. One formalism links perception, prediction, action and uncertainty. It also supplies an agent-side counterpart to risk assigned by an observer (see A Suggested Formula below).

• Limitation. The formulation applies to "systems such as coupled pendulums that one would not readily recognize as autonomous", so it needs further criteria for agency.34 Critics argue that Markov blankets cannot fix a single principled boundary between agent and world, because that boundary depends on modelling choices. They propose treating the blankets as instrumental constructs.35 Preferences, finally, enter as model parameters, which should not be read as conscious desires.

Biosemiotic Risk

• Definition. Risk is the possibility that signs, or their absence, lead an agent into conditions that matter negatively for its way of life.

• Scope. Biosemiotics holds that "the semiosic–non-semiosic distinction is coextensive with the life–nonlife distinction".36 Every living being interprets signs, so every living being can misread them.

• Proto-signs. Primitive organisms cannot classify or track objects. Their sign use therefore relies on "proto-signs that directly regulate activities of agents without being associated with objects". Object recognition became possible later, in eukaryotes that developed rewritable epigenetic memory as well as sensory and effector capacities.37 Water chemistry, light or pressure can thus act as risk cues without becoming representations.

• Priors limit learning. If an animal's ancestors never met aerial predators, its prior expectation of danger from above may be zero, so "it might never learn to be wary of 'open sky'".38 A risk outside an agent's world of signs can still harm it.

• Ecological traps. Organisms rely on cues that, over evolutionary time, correlated with survival and reproduction. In environments that humans alter suddenly, "formerly reliable cues might no longer be associated with adaptive outcomes", and organisms become trapped.39 Insects that use polarised light to find water surfaces, for example, read glass buildings, asphalt roads and some cars as water. They then defend territories there and lay eggs.40 See also ecological traps.

  

  ^{Habitat choice by cues in a normal environment (a) and in an ecological trap (b) (Fig. 1 in the source).39}

• Eel example: conspecific odour. Glass eels of the American eel (Anguilla rostrata) were attracted to washings of the previous year's elvers. They "may use the odour of the previous year class as information to guide migration". The authors consider decreased predation as well as habitat quality as possible reasons for such attraction.41

• Eel example: light. Of migrating American eels that met a 216 m floating LED array, 39.5% changed course consistent with lateral deflection. Most, 76.7%, dived 4 m or more, "consistent with seeking darker passage under the light field".42 The eels treated light as a sign to avoid, mostly by changing depth.

• Strength. The approach takes organism-relative meaning seriously. It explains why the same condition is a risk cue for one being and noise for another.

• Limitation. On this account semiosis coincides with life,36 so extending biosemiotic risk to machines, ecosystems or nonliving systems needs further argument.

A Suggested Formula

This section is the note's own proposal. It states the working definition for any bearer of risk, then relates the three formal accounts and the Red List to it.

• Calculative baseline. Infrastructure risk is often written as $R = P \times C$: the probability of an adverse event multiplied by its consequences. The same source frames risk as danger "to people or what we value", which leaves the evaluator human.43 A more general definition treats risk as a set of triplets $\{\langle s_i, p_i, x_i \rangle\}$, where $s_i$ is a scenario, $p_i$ its likelihood and $x_i$ its consequence. It rejects the product form because "the probability times consequence viewpoint would equate a low-probability high-damage scenario with a high-probability low-damage scenario". The analysis asks how the future will turn out "if we undertake a certain course of action (or inaction)", which corresponds to $\pi$ below.44

• Agent-relative risk. For a bearer $a$ that follows a course of action $\pi$ under disturbances $D$:

  $$R_a(\pi, \tau) = P\big(x_a(t+\tau) \notin V_a \mid x_a(t), \pi, D\big)$$
  • $x_a(t)$: the state, at time $t$, of the variables that $a$ regulates (Ashby's essential variables $E$).29
  • $V_a$: the set of states in which $a$ remains viable or achieves what it acts towards (Ashby's permissible set $\eta$; the inside of the viability boundary).27^, 29
  • $\pi$: the course of action, including doing nothing.
  • $D$: disturbances from the environment, including other agents.
  • $\tau$: the time horizon.
  • $P$: a probability assigned by an observer or a model. The bearer itself need not compute it.

• Assessable risk. Let $T_a$ be the temporal reach of the agent's cognitive light cone.28 The agent can assess $R_a$ only for $\tau \leq T_a$. Risks with longer horizons can still harm it.

• Graded harm. Viability norms are graded, distinguishing "good, neutral, and bad ways" of self-maintenance.27 A graded version replaces leaving $V_a$ with a harm function $h_a(x) \geq 0$, which is zero in preferred states and rises towards the viability boundary:

  $$H_a(\pi, \tau) = \mathbb{E}\big[h_a\big(x_a(t+\tau)\big) \mid x_a(t), \pi, D\big]$$

  The binary form above is the special case in which $h_a$ equals 1 outside $V_a$ and 0 inside it.

• Red List as a special case. Let $a$ be a species population and $V_a$ its persistence in the wild. With $\tau$ set to ten years or three generations, whichever is longer (up to 100 years), criterion E classes a taxon as Critically Endangered when $R_a \geq 0.5$.15 A species need not assess anything for this to work, because the formula measures risk to a bearer, while assessment is a separate capacity (see Limit above).

• Link to cybernetics. Requisite variety bounds what regulation can achieve. When the variety of an agent's responses $V_R$ falls short of the variety of disturbances $V_D$, outcomes spread by at least $V_D - V_R$. $R_a$ then rises unless that spread stays within $V_a$.29

• Link to active inference. The active-inference risk term is the agent-side counterpart. It replaces the observer's probability of leaving $V_a$ with the divergence of predicted outcomes from preferred ones, computed under the agent's own generative model.32 Preferences $C$ that give low probability to states outside $V_a$ align the two measures.

• Link to biosemiotics. Ecological traps show where $R_a$ and the agent's own assessment part ways. Cues that once indicated states inside $V_a$ now lead outside it, so the agent's estimate of risk falls while $R_a$ rises.39

• Eel reading. For a silver eel at a hydropower intake, $x_a$ includes body integrity, $V_a$ excludes injury, and $D$ includes turbine flows as well as artificial light. Behavioural guidance tries to lower $R_a$ by presenting signals the eel already avoids.31^, 42

Questions for a More-than-Human Risk Assessment

Terms

This note extends IPCC usage of exposure and vulnerability, which covers people as well as assets, to any agent.13

References

Lupton, Deborah. Risk. London: Routledge, 1999.

Lupton, Deborah, ed. Risk and Sociocultural Theory: New Directions and Perspectives. Cambridge: Cambridge University Press, 1999.

Notes

---

Footnotes

1. Dequech, David. “Uncertainty: A Typology and Refinements of Existing Concepts.” Journal of Economic Issues 45, no. 3 (2011): 621–40. https://doi.org/10.2753/JEI0021-3624450306.˄

2. Wynne, Brian. “Uncertainty and Environmental Learning: Reconceiving Science and Policy in the Preventive Paradigm.” Global Environmental Change 2, no. 2 (1992): 111–27. https://doi.org/10.1016/0959-3780(92)90017-2.˄

3. Aven, Terje. “On How to Deal with Deep Uncertainties in a Risk Assessment and Management Context.” Risk Analysis 33, no. 12 (2013): 2082–91. https://doi.org/10.1111/risa.12067.˄

4. Walker, Warren E., Poul Harremoës, Jan Rotmans, Jeroen P. van der Sluijs, Marjolein B. A. van Asselt, Peter P. M. Janssen, and Martin P. Krayer von Krauss. “Defining Uncertainty: A Conceptual Basis for Uncertainty Management in Model-Based Decision Support.” Integrated Assessment 4, no. 1 (2003): 5–17. https://doi.org/10.1076/iaij.4.1.5.16466.˄

5. Dewulf, Art, and Robbert Biesbroek. “Nine Lives of Uncertainty in Decision-Making: Strategies for Dealing with Uncertainty in Environmental Governance.” Policy and Society 37, no. 4 (2018): 441–58. https://doi.org/10.1080/14494035.2018.1504484.˄

6. Brugnach, Marcela, Art Dewulf, Claudia Pahl-Wostl, and Tharsi Taillieu. “Toward a Relational Concept of Uncertainty: About Knowing Too Little, Knowing Too Differently, and Accepting Not to Know.” Ecology and Society 13, no. 2 (2008): art30. https://doi.org/10.5751/ES-02616-130230.˄

7. Douglas, Mary, and Aaron Wildavsky. Risk and Culture: An Essay on the Selection of Technological and Environmental Dangers. Berkeley: Sage, 1983.˄

8. Rappaport, Roy A. “Toward Postmodern Risk Analysis.” Risk Analysis 8, no. 2 (1988): 189–91. https://doi.org/10.1111/j.1539-6924.1988.tb01169.x.˄

9. Ulturgasheva, Olga, and Mally Stelmaszyk. “Embracing Uncertainty: Porous and Actionable Responses to Climate Change at the Borders of Indigenous and Scientific Expertise(s) in Siberia.” Journal of the Royal Anthropological Institute 31, no. 1 (2025): 63–81. https://doi.org/10.1111/1467-9655.14163.˄

10. Böhm, Gisela, and Carmen Tanner. “Environmental Risk Perception.” In Environmental Psychology: An Introduction, 2nd ed., edited by Linda Steg and Judith I. M. de Groot, 15–25. Hoboken: Wiley-Blackwell, 2019. https://doi.org/10.1002/9781119241072.ch2.˄

11. Beck, Ulrich. Risk Society: Towards a New Modernity. Translated by Mark Ritter. 1986; London: SAGE, 1992.˄

12. Pickering, Andrew. “Asian Eels and Global Warming: A Posthumanist Perspective on Society and the Environment.” Ethics & the Environment 10, no. 2 (2005): 29–43. https://doi.org/10.1353/een.2005.0023.˄

13. Rockström, Johan, Joyeeta Gupta, Dahe Qin, Steven J. Lade, Jesse F. Abrams, Lauren S. Andersen, David I. Armstrong McKay, et al. “Safe and Just Earth System Boundaries.” Nature 619, no. 7968 (2023): 102–11. https://doi.org/10.1038/s41586-023-06083-8.˄

14. New, Mark, Diana Reckien, David Viner, Carolina Adler, So-Min Cheong, Cecilia Conde, Andrew J. Constable, et al. “Decision-Making Options for Managing Risk.” In Climate Change 2022: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, edited by Hans-Otto Pörtner, Debra C. Roberts, Melinda Tignor, Elvira S. Poloczanska, Katja Mintenbeck, Andrés Alegría, Marlies Craig, et al., 2539–654. Cambridge: Cambridge University Press, 2022. https://doi.org/10.1017/9781009325844.026.˄

15. IUCN Species Survival Commission. IUCN Red List Categories and Criteria: Version 3.1. 2001. 2nd ed. Gland: International Union for Conservation of Nature, 2012.˄

16. Bostrom, Nick. “Existential Risks: Analyzing Human Extinction Scenarios and Related Hazards.” Journal of Evolution and Technology 9, no. 1 (2002). https://www.jetpress.org/volume9/risks.html.˄

17. Kareiva, Peter M., and Valerie Carranza. “Existential Risk Due to Ecosystem Collapse: Nature Strikes Back.” Futures 102 (2018): 39–50. https://doi.org/10.1016/j.futures.2018.01.001.˄

18. Liu, Hin-Yan. “Anthropocene Liminality: Lurching towards the Synthetic Age in the Twilight of Gaia.” Law, Innovation and Technology 17, no. 2 (2025): 873–907. https://doi.org/10.1080/17579961.2025.2593776.˄

19. Gordon, Bonnie J., and Stanislav Roudavski. “More-than-Human Infrastructure for Just Resilience: Learning from, Working with, and Designing for Bald Cypress Trees (Taxodium distichum) in the Mississippi River Delta.” Global Environment 14, no. 3 (2021): 442–74. https://doi.org/10.3197/ge.2021.140302.˄

20. Lima, Steven L., and Lawrence M. Dill. “Behavioral Decisions Made under the Risk of Predation: A Review and Prospectus.” Canadian Journal of Zoology 68, no. 4 (1990): 619–40. https://doi.org/10.1139/z90-092.˄

21. Gaynor, Kaitlyn M., Joel S. Brown, Arthur D. Middleton, Mary E. Power, and Justin S. Brashares. “Landscapes of Fear: Spatial Patterns of Risk Perception and Response.” Trends in Ecology & Evolution 34, no. 4 (2019): 355–68. https://doi.org/10.1016/j.tree.2019.01.004.˄

22. Suraci, Justin P., Michael Clinchy, Liana Y. Zanette, and Christopher C. Wilmers. “Fear of Humans as Apex Predators Has Landscape-Scale Impacts from Mountain Lions to Mice.” Ecology Letters 22, no. 10 (2019): 1578–86. https://doi.org/10.1111/ele.13344.˄

23. Selbach, Christian, Loïc Marchant, and Kim N. Mouritsen. “Mussel Memory: Can Bivalves Learn to Fear Parasites?” Royal Society Open Science 9, no. 1 (2022): 211774. https://doi.org/10.1098/rsos.211774.˄

24. Plumwood, Val. “Human Vulnerability and the Experience of Being Prey.” Quadrant 39, no. 3 (1995): 29–34.˄

25. Dooren, Thom van. Flight Ways: Life and Loss at the Edge of Extinction. New York: Columbia University Press, 2014.˄

26. Whyte, Kyle. “Our Ancestors’ Dystopia Now: Indigenous Conservation and the Anthropocene.” In The Routledge Companion to the Environmental Humanities, edited by Ursula K. Heise, Jon Christensen, and Michelle Niemann, 222–31. Abingdon: Routledge, 2017. https://doi.org/10.4324/9781315766355-32.˄

27. Di Paolo, Ezequiel A. “Autopoiesis, Adaptivity, Teleology, Agency.” Phenomenology and the Cognitive Sciences 4, no. 4 (2005): 429–52. https://doi.org/10.1007/s11097-005-9002-y.˄

28. Levin, Michael. “The Computational Boundary of a ‘Self’: Developmental Bioelectricity Drives Multicellularity and Scale-Free Cognition.” Frontiers in Psychology 10 (2019): 2688. https://doi.org/10.3389/fpsyg.2019.02688.˄

29. Ashby, W. Ross. An Introduction to Cybernetics. London: Chapman & Hall, 1956.˄

30. Conant, Roger C., and W. Ross Ashby. “Every Good Regulator of a System Must Be a Model of That System.” International Journal of Systems Science 1, no. 2 (1970): 89–97. https://doi.org/10.1080/00207727008920220.˄

31. Piper, Adam T., Costantino Manes, Fabio Siniscalchi, Andrea Marion, Rosalind M. Wright, and Paul S. Kemp. “Response of Seaward-Migrating European Eel (Anguilla anguilla) to Manipulated Flow Fields.” Proceedings of the Royal Society B: Biological Sciences 282, no. 1811 (2015): 20151098. https://doi.org/10.1098/rspb.2015.1098.˄

32. Parr, Thomas, Giovanni Pezzulo, and Karl J. Friston. Active Inference: The Free Energy Principle in Mind, Brain, and Behavior. Cambridge, MA: MIT Press, 2022.˄

33. Parr, Thomas, and Karl J. Friston. “Generalised Free Energy and Active Inference.” Biological Cybernetics 113, nos. 5–6 (2019): 495–513. https://doi.org/10.1007/s00422-019-00805-w.˄

34. Kirchhoff, Michael, Thomas Parr, Ensor Rafael Palacios, Karl J. Friston, and Julian Kiverstein. “The Markov Blankets of Life: Autonomy, Active Inference and the Free Energy Principle.” Journal of The Royal Society Interface 15, no. 138 (2018): 20170792. https://doi.org/10.1098/rsif.2017.0792.˄

35. Bruineberg, Jelle, Krzysztof Dołęga, Joe Dewhurst, and Manuel Baltieri. “The Emperor’s New Markov Blankets.” Behavioral and Brain Sciences 45 (2022): e183. https://doi.org/10.1017/S0140525X21002351.˄

36. Kull, Kalevi, Terrence W. Deacon, Claus Emmeche, Jesper Hoffmeyer, and Frederik Stjernfelt. “Theses on Biosemiotics: Prolegomena to a Theoretical Biology.” Biological Theory 4, no. 2 (2009): 167–73. https://doi.org/10.1162/biot.2009.4.2.167.˄

37. Sharov, Alexei A. “Evolutionary Biosemiotics and Multilevel Construction Networks.” Biosemiotics 9, no. 3 (2016): 399–416. https://doi.org/10.1007/s12304-016-9269-0.˄

38. Dall, Sasha R. X., Luc-Alain Giraldeau, Ola Olsson, John M. McNamara, and David W. Stephens. “Information and Its Use by Animals in Evolutionary Ecology.” Trends in Ecology & Evolution 20, no. 4 (2005): 187–93. https://doi.org/10.1016/j.tree.2005.01.010.˄

39. Schlaepfer, Martin A., Michael C. Runge, and Paul W. Sherman. “Ecological and Evolutionary Traps.” Trends in Ecology & Evolution 17, no. 10 (2002): 474–80. https://doi.org/10.1016/s0169-5347(02)02580-6.˄

40. Van Dyck, Hans. “Changing Organisms in Rapidly Changing Anthropogenic Landscapes: The Significance of the ‘Umwelt’-Concept and Functional Habitat for Animal Conservation.” Evolutionary Applications 5, no. 2 (2012): 144–53. https://doi.org/10.1111/j.1752-4571.2011.00230.x.˄

41. Galbraith, Heather S., Carrie J. Blakeslee, Andrew K. Schmucker, Nicholas S. Johnson, Michael J. Hansen, and Weiming Li. “Donor Life Stage Influences Juvenile American Eel Anguilla rostrata Attraction to Conspecific Chemical Cues.” Journal of Fish Biology 90, no. 1 (2017): 384–95. https://doi.org/10.1111/jfb.13190.˄

42. Elvidge, Chris K., M. E. Cole MacLeod, Daniel Hatin, Alexandre Guindon, René Dion, Scott Schlueter, Justin Ecret, et al. “River-Scale Evaluation of a Floating LED Light Array for Behavioral Guidance of American Eel.” River Research and Applications 42, no. 8 (2026): 1623–37. https://doi.org/10.1002/rra.70148.˄

43. Little, Richard G. “Managing the Risk of Cascading Failure in Complex Urban Infrastructures.” In Disrupted Cities: When Infrastructure Fails, edited by Stephen Graham, 27–39. Abingdon: Routledge, 2010.˄

44. Kaplan, Stanley, and B. John Garrick. “On the Quantitative Definition of Risk.” Risk Analysis 1, no. 1 (1981): 11–27. https://doi.org/10.1111/j.1539-6924.1981.tb01350.x.˄

---

Backlinks

• Housing
• Success
• Uncertainty
• Sentience Book Project
• Ignorance
• Need