by Mike O’Brien
What a week it has been. I’m not referring to military outrages, legal bombshells or pop-cultural bombshells. Rather, I’m referring to the dozens of intensive (and intensely rewarding!) hours I spent catching up on my preferred corner of academic research: the empirical investigation of animal normativity. Big things are happening in this domain. Big things have been happening there for decades, but the pace has noticeably increased in the last two years, at least judging by the output of the authors I tend to follow.
Some of my readers may know that I am particularly interested in the work of Kristin Andrews, currently at York University in Toronto. I have covered some publications of hers in previous columns, most recently 2022’s “A pluralistic framework for the psychology of norms“, co-written with Evan Westra. Since then, no fewer than nine publications have been added to Andrews’ website, many co-authored with other movers and shakers in the burgeoning animal normativity scene. In addition to illustrating the current state of the field, the historical references in these recent publications (if I can call the 1970s and 1990s “historical” without sending a chill up the spines of my peers and elders) also trace the long trajectory of de-anthropocentrizing projects in cognitive and behavioural sciences. A particularly interesting antecedent is 1990’s “Towards a neurobiological theory of consciousness” by Francis Crick (!) and Christof Koch, which called for a program of research into consciousness that presupposed a neuronal rather than linguistic basis for conscious phenomena. A fortuitous proposal, in retrospect, accompanied by some rather interesting specific hypotheses about the underlying neuronal mechanics. Koch’s confidence in the material tractability of consciousness recently cost him a case of wine (presumably now sitting in David Chalmers’ cellar), although he should not be faulted for the mind-sharpening practice of attaching stakes to one’s bets.
To recap my previous coverage of Andrews’ work: In “A pluralistic framework…“, Andrews and Westra sketched out a conceptual toolkit for a research program that could investigate normativity in non-human animals. The main thrust of this project is to remove heavily concept-laden and human-specific definitions and criteria, and in their stead provide minimal, instrumentally serviceable tools that can be applied to a wide variety of animal behaviours. Rather than assuming that humanity is the apex of some evolutionary pyramid, Andrews’ and Westra’s approach accommodates the possibility that normativity emerges across the natural world, in forms that need not relate to the human varieties that we know by familiarity and introspection. Most of the recent publications are elaborations of this approach, putting some tentative meat on the bones of the framework. I will put to one side the works that deal with comparisons between AI and animal consciousness, although they are quite interesting and intimately tied to methodological questions about identifying conscious agents through behaviour. (I may revisit this area in a future piece about the work of Walter Veit, who is doing some promising research on non-human consciousness, and who hosts a podcast featuring very interesting conversations with other researchers doing related work. I strongly recommend checking it out.)
In 2022’s “How mindshaping and social maintenance can support shared intentions in great apes“, Andrews and co-author Dennis Papadopoulos argue that the observed absence among apes of certain behaviours has been misinterpreted as evidence that these animals cannot form “shared intentions”, i.e. a joint commitment to some goal, with participants having standing to rebuke other participants for failure. They propose that the missing behaviours, such as sharing rewards according to the size of a helper’s contribution, or protesting inequitable sharing, may not occur in these ape societies because of a hierarchical structure in which equity is not expected, and in which protest against dominant members does not usually occur. Rebuke and protest may also take subtler and more temporally diffuse forms, such as decreased likelihood to cooperate in the future, or diminished positive interactions, that escape the attention of researchers who are looking for discrete and immediate signs. Such a narrow window of observation would also miss many human forms of dissent and rebuke, especially in the context of long-term, complex and hierarchical relations wherein direct and immediate reactions may be rare and costly.
“Mindshaping” in the context of this paper means the updating of expectations about the future, without the need for any propositional attitudes or explicit beliefs about the behaviour of other group members, much less a theory of mind about their reasons for behaviour. Such updating is less cognitively demanding than more elaborate mental modelling, and widens the range of species that may be capable of shared intentional behaviour. Animals need not have the cognitive capacity to understand a theory of cooperation in order to cooperate successfully. This is an example of Andrews’ and Westra’s approach in action: seeking minimally complex and demanding explanations for behaviour, such that candidate species are not excluded due to unproven assumptions about what capacities they ought to have in order to do things they may already be doing.
In 2023’s “Rule-ish patterns in the psychology of norms“, Andrews and Westra criticize Cecilia Heyes’ proposal (in “Rethinking norm psychology“) of 3 types of normative behaviour: compliance, enforcement, and commentary. The criticism is that these candidate types are too rule-dependent and thus too psychologically demanding, excluding many candidate animals that exhibit normative behaviour but lack the capacity for complex representational cognition. Andrews and Westra argue that rules are an explanatory construct, not a feature of the world to be explained, whereas normative behaviour itself is what Daniel Dennett called a “real pattern”, manifest to our pre-theoretical observation of the world. (I have not read Heyes yet, so I can’t comment on the accuracy of this critique. My reading docket is approaching double-digit multiples of my expected lifetime, so there will be gaps.)
Continuing the project of knocking down a priori exclusions of animals from the normativity picture, 2024’s ““All animal are conscious”: Shifting the null hypothesis” tackles problems with what Andrews calls the “marker approach”. This consists of using markers (behaviours, anatomical features) as a proxy for the presence of some property, like consciousness. Andrews notes that some commonly accepted markers of consciousness can be found in a creature as simple as C. elegans, a nematode possessing about three hundred neurons (contrast this to a hundred thousand neurons in the brain of a fruit fly, or eighty-six billion neurons in the brain of a human). So, should we accept that the nematode is conscious? Or reject the validity of the specific markers possessed by it? Or reject the marker approach altogether? Andrews proposes that we drop the marker approach for answering the “distribution question” of consciousness, i.e. which animals are conscious, and instead adopt the working assumption that all animals are conscious. The marker approach is more useful for answering the “dimensions question” of what kinds of consciousness there are. (Andrews points to Jonathan Birch et al.’s 2021 “Review of the evidence of sentience in decapod crustaceans and cephalopod molluscs“, which resulted in the extension of UK welfare laws to these groups of animals, as a model of best practice for the markers approach.) Since no empirically or logically “secure” theory exists for excluding the possibility of animal consciousness a priori, we may as well adopt a “theory-light” approach and see what nature has to show us.
Andrews further argues that if we accept the marker approach for the distribution question, and have no negative markers to exclude a candidate from consideration, and discover new markers while old markers are never completely discarded, we will inevitably end up with increased confidence that any initially worthwhile candidate species is conscious. So the marker approach is neither a reliable positive test (as the methodology drifts towards positive conclusions by itself) nor a reliable negative test (as we have no negative markers). Further elaborations are made on the distinctions between “initial” markers (intuitively, pre-theoretically obvious) and “derived” markers (observable, generalizable features associated with initial markers), with attention given to potential problems such as the weakening of evidentiary force as derived markers are distanced from initial markers by successive associative leaps.
In addition to these arguments for abandoning the marker approach for the distribution question, Andrews makes a positive case for the “all animals are conscious” null hypothesis on two fronts. First, there are pragmatic benefits for researchers to adopt this view, since remaining agnostic or denying consciousness could cause them to ignore or omit important aspects of animal behaviour from their experimental designs and observations. Unaccounted-for variability in the reactions of conscious subjects to their environment can produce confounding artifacts in experimental results. Second, it is generally wise to study simpler instances of a phenomenon, as more complex instances may be bundled with other dynamics and variables that are not essential to the object of study. In the case of consciousness, by studying humans and closely related species we risk mistaking evolutionarily recent and contingent features for essential and paradigmatic markers of consciousness.
The latest milestones of Andrews et al.’s project is 2024’s “In search of animal normativity: A framework for studying social norms in non-human animals” by Westra et al., and “Human and non-human norms: A dimensional framework” by Andrews, Fitzpatrick and Westra (they may be superseded in recency at any minute; these people are busy). In the first paper, the authors try to lay out a research program for animal normativity inspired by the progress made in the field of animal culture since the 1990s. Following that program’s shift from heavily psychological and theory-laden approaches to more empirically tractable definitions of culture, they base this normativity research program on the core notion of “normative regularity”, defined as “socially maintained patterns of behavioural conformity within a community”. They point out that little consensus exists about the underlying psychological mechanisms of human normativity, so looking for those mechanisms in animals is not a promising approach. Also, theories that rely on representations of “is” and “ought” states face difficulties in linking such mental representations to observable behaviour. Further, because of the group setting in which normativity is manifest, studying individual capacities (as might be done in studying animals’ number sense or theory of mind) is not adequate.
The authors go into great detail, that I will not attempt to summarize here, about how the successful pluralistic approach in animal culture research should be emulated, listing three main lessons from that program: First, the need for working conceptions that allow researchers to refer to patterns of interest and are neutral with regard to underlying psychological mechanisms, nor burdened by assumptions about cognitive capacity or evolutionary history. Second, the ability to capture paradigmatic human norms without being human-centric. Third, the need to develop a taxonomy with finer-grained distinctions between kinds of norms and levels of complexity. (The most complex kind of normative regularities being “recursive normative regularities”. More on recursion later… ). The authors point out that the project they describe is one of conceptual engineering, to develop useful tools for researching the phenomena of interest, and not conceptual analysis to explain existing folk intuitions about norms.
From there the authors propose a number of candidate definitions, candidate variables for study, and decomposed and fine-grained expansions on the core notions of “patterns of behavioural conformity”, “social maintenance” and “community”. They also discuss the benefits of this theory-light pluralistic approach for animal research and for conservation efforts, as well as for enriching the evolutionary understanding of human norm psychology. Again, it is too dense and detailed to summarize, so I recommend that interested readers consult the paper for themselves (all but one of the recent publications on Andrews’ website is open-access).
“Human and non-human norms…” is also largely about making concrete proposals for implementing the framework developed by Andrews, Westra and their fellow travellers. Sticking with the working definition of “socially-maintained patterns of behavioural conformity within a community”, this paper proposes several “dimensions of variability” at the level of individual psychology (rule-following, behavioural understanding, collective agency, and motivation), and at the level of social interaction (pedagogy and punishment). These dimensions are divided into levels of complexity, from the most basic and non-psychologically-demanding through to the most complex and usually human-unique. For example, “rule-following” can be as basic as behavioural patterns emerging from shared biology, or as complex as public debates about rules. This paper is even less summarizable than the preceding one, because it more an example of concrete application and less an argument for the underlying approach. By elaborating a scheme for norm investigation that is applicable to animals ranging from (perhaps) nematodes and adult humans, it marks an important step in the maturation of the research program proposed scarcely two years ago.
I realize this kind of methodological work can be rather dull if you are not already invested in the projects that they instrumentally serve, and familiar with the methodological weaknesses they seek to address. Let me finish this round-up with something closer to what got me excited about this line of research years ago.
2023’s “Humans, the norm-breakers” is a commentary by Andrews of Victor Kumar and Richmond Campbell’s book “A Better Ape“, and I had previously skipped over it because I had mistaken it for a book review, rather than a philosophy article occasioned by a book. It turns out that this article addresses quite directly a question that has been on my mind for years, after posing it to Andrews at a talk and being told that it was as yet unanswered. That question is: do great apes, who show strong evidence of having normative behaviour and psychology, show any signs of having norms about norms? That is to say, when a chimpanzee is faced with a situation in which two conflicting norms apply, does she have access to, say, a rule for sorting norms by priority or importance?
This question arises in the critique of Kumar and Richmond’s argument that norms provided humanity with a more precise and flexible means of navigating quickly changing societies, and of their argument that norms are rules put into language, which makes language a prerequisite for normativity. Andrews argues that this is an over-specified and overly demanding picture of norms, and that norms mostly serve a conservative rather than innovative function, limiting rather than enhancing flexibility in changing conditions. What is required, for the kind of flexibility that Kumar and Richmond attribute to norms, is a means of breaking certain norms, without breaking the system of norms; allowing innovators to violate norms without punishment, and allowing some norms to be revised or discarded while others are conserved. In short, norms about norms. Language, while not necessary for normativity simplicitur, is necessary for making norms the object of debate and revision within a community, allowing for a use/mention distinction with regard to norms. The combination of normativity (which Andrews and others argue pre-dates language) and language allows humans to create social practices of giving and considering reasons, and thus accumulate a normative system that is (at least potentially) adaptive to change.
Andrews states that the existing literature still shows no evidence of such meta-normative capacities and practices in non-human animals. Such a gulf between our species and all others is why I find stories like the rebooted “Planet of the Apes” series so compelling. But we are still in the early days of figuring out how to investigate these apparent quantum gaps, and I have some hunches yet about how they may be bridged…
For links to Dr. Andrews’ publications:
https://www.kristinandrews.org/publications
For my previous discussion of Dr. Andrews’ work:
https://3quarksdaily.com/3quarksdaily/2023/05/how-ought-we-think-about-ought-thoughts.html