Do bacteria think? 10 Dec 2007 Let’s suppose there is a game, say, baseball. This game is named and described for the ways that adult humans with bats, balls, and fields, behave normatively, as written up in an authoritative manual. Everybody knows what baseball is, or can point to an example of it. Along comes someone, however, who notes that there is a formal resemblance between baseball and what some ants do in some hitherto undiscovered nest. So, they start to call it “ant baseball”. So far, no harm. Then someone else comes along and starts making inferences about what ants do in terms of the rules of the game as laid down in the manual. That would be at best, overextending the analogy, right? At worst it is a fallacy. Now replace “baseball” with “cognition” and “ants” with “bacteria. Recently there has been a lot of discussion of the cognitive capacities of microbes, particularly eubacteria, when they live in communities of many species (often referred to as biofilms). One of the common themes is that bacteria have a kind of cognitive capacity based on their ability to send “signals” to each other that affects the behaviour of the community. Signal transduction is a well known, if as yet not completely understood, process. It is how neurons and other cells in our bodies affect each other by passing chemicals between themselves, and activating processes in response inside each cell. It is also how bacteria “communicate” in communities. Pamela Lyon, of the University of Adelaide and the ANU, has been arguing for some time that this makes what bacteria do a form of cognition, and she has argued that cognition is as basic a biological property as respiration. I think she’s mistaking ant baseball for real baseball. Now I very much like Pamela, and she does this in a sophisticated and technical manner that is decidedly not stupid, so rather than riff on her work, I’d like to consider it as an instance of a more general matter in biology: that of seeing the world in human terms, or, as we call it, anthropomorphising. Is it wrong and why? Friends with forbearance and hapless victims who know me know that I take an extremely sparse approach to the ontology of biology. I reject the notion that there is information in, say, genes. I reject the notion that functions exist outside our characterisations of biological systems. I deny that there is “intelligence” or “problem solving” in natural selection; and so on. In short, I take the line that unless there is a compelling need to use categories designed to deal with human beings at the gross level in biology, that we shouldn’t, or if we do, we need to know this is only an “honorary” usage, and nothing should be inferred from it. Anthropomorphism has a long history in biology. The entire tradition of ascribing purposes in biology – known as teleology – is deeply seated in the western tradition. Aristotle used it (his “final cause”) and almost everyone has used it since. One of the radical departures from this is, of course, evolution by natural selection. “Goals” and “purposes” are determined post hoc. Telos evaporates, except in such nervous systems as are able to model their world and determine a course of action before acting (and even there, the goals are the result of a process of prior selection of models). Francis Bacon triggered off a revolution when he declared that final causes were useless in science, except when talking about human affairs. It seems that 600 years later we are still learning that lesson: the final cause rather corrupts than advances the sciences, except such as have to do with human action [Novum organon, Bk II, aph 2] So why do we insist on using it and like concepts, developed to speak of human affairs, when we approach biological objects of a very different kind? Why talk about “proofreading” when speaking of genes, or the “intentionality” of states of organisms, or, as in Pamela’s case, the intelligence of bacteria? It is at best a kind of useful rhetoric, the way Dawkins spoke of “selfish genes”, designed to shake loose preconceptions about how these biological objects behave. But in Dawkins’ case, there was an excuse – Maynard Smith and others had applied the mathematical theory of games to genetics, and the anthropomorphism there was incidental to the substantive point that the logic of games was a useful mathematical tool for discussing the evolution of genes. Still, a number of, I have to say, philosophers (like Mary Midgley) took this as a serious claim – that genes were interested only in themselves. Sensible folk knew this was a way of speaking, not a substantial claim, but the waters were muddied thereafter, and now we have to disentangle the real sense of “selfishness” or “altruism”, as applied to motives of human agents, from the analogical sense applied to genes, to students in undergraduate courses. Tedious, but used properly it’s a teaching opportunity. Back in 1990, a fellow named Jonathon Schull tried to argue that populations were “problem solvers”, and that their evolution was a kind of learning experience. It was a fancy way to claim that populations reach and remain at an adaptive peak, but the rhetoric was hard to disentangle from the actual claim. Now we are faced with philosophical arguments that ensembles of bacterial species in biofilm communities actually do cognise. How much is rhetoric and how much is substantial. At a conference in 2006 in Exeter, Pamela and others, including bacteriologist James Shapiro, argued that this was not merely a rhetorical flourish, but a fact. I think they have mistaken the metaphor for the model, as many have done before them in terms of genes, functions, and so on. If there is some formal resemblance between what happens in neural networks when they solve problems, and what happens in bacteria when they are faced with ecological challenges (another metaphor), that doesn’t mean either that they have any other similarity that is significant for cognition. It is no more important than selfishness is for genes. Possibly less so. As I asked Shapiro at that conference, what gets lost if we simply use unloaded terms and not terms like “knowledge”, “information” and “problem” to describe what bacterial communities do? I say nothing. Pamela and Shapiro say, lots. But I am unconvinced. We need to administer a kind of intellectual enema to rid ourselves of anthropic projection in biology. It seems to me that we’d do a lot better if we taught biologists to distinguish between language that helps us think (for we evolved to think in terms of agency and intelligence) and what the properties of the biological objects themselves are separate to how we describe them. There’s a fallacy in play here, an Intentional Fallacy we might call it. Genes do not get “proofread”, they get modified by molecular ensembles so that mismatches are changed to match the most methylated (probably the oldest) strand of DNA. Bacteria do not solve problems or learn; they adapt their interactions to deal with changes in the environment, with machinery (another metaphor) that evolved for that purpose (another metaphor). We understand the biological world best in its own terms, and if that means we have to uneducate our language first, then so be it. Evolution General Science
Epistemology Does philosophy generate knowledge? 2 Sep 20122 Sep 2012 So Larry has responded. Go read it. I’ll wait…. Back? Good. Let me address some of the points there. Not all of them, because most of them I have already addressed in previous posts. I’ll link them at the end of this one. But the most important ones. The first… Read More
Biology Zimmer’s evolving blog post on X-woman 28 Mar 2010 Carl Zimmer has a continually updated blog post on the mtDNA of “X-Woman”, which is being informed by his readers and experts. It’s a useful antidote to hasty and inaccurate reportage in the MSM, done well by a very good journalist. Read More
Evolution Changing minds 1 Jan 2008 The Grauniad has a puff piece on what has changed the minds of “the intellectual elite”. It seems Alan Alda is one of them in virtue of his screen roles… still I have to like his comment: Until I was 20 I was sure there was a being who could… Read More
I agree with you that to say that bacteria think is a stretch. But the reason it’s a stretch is because we don’t even understand how we think in any meaningfully systematic way, so any comparison is meaningless. “Proofreading,” on the other hand, is a different matter, in my opinion. DNA is typically thought of as “proofread” first by DNA polymerase itself — if a bad nucleotide is incorporated into the DNA, DNA polymerase literally stalls briefly and then excises the bad base. Now, the stalling and excision are, of course, entirely physical processes with no intention behind them. But the metaphor, I think, does convey information to somebody who doesn’t have time to study the subtle kinetics involved, provided that they understand that there are subtle kinetics driving the process. The metaphor is useful because it conveys a sense that the enzyme stops, “erases” what it just did, and continues on — which is pretty much what does happen. So I do agree with you that language can obscure understanding. But I don’t think that’s always the case.
As a behaviorist (yes, there are still behaviorists), I say with all sincerity “good luck”. Anthropomorphism and the resultant circular logic is a problem when we look at human thinking, let alone non-human animal behavior. I suppose it was fine when we really didn’t have a chance to understand what was actually going on… but now we have a real shot at real explanations, and anthropomorphism, circular reasoning, and the very concept of “mind” simply get in the way. Good luck.
Dunno. I read about how the bacteria that come up in the raw oil and get pumped long distances start (re?)assembling themselves into increasingly complicated structures on the inside of the pipes, and I wonder — what sort of structure had they created way down there in the warm dark, with some tens of millions of years available? And what are they doing now pulled up and scattered around in the cold bright air? Slow thoughts, at best, I guess, but if it’s thoughts per generation, they may still be learning faster than we do.
I’m not sure how I see claiming that genes store information is anthropomorphizing. From a computational perspective, our DNA clearly contains the set of instructions responsible for building our bodies. Just because no person was responsible for poking our genes into a bunch of useful patterns doesn’t mean the term doesn’t apply. I’ll give you that Computer Scientists tend to anthropomorphise quite a bit, tho.
Phill, you just did exactly what I was talking about. DNA contains no “instructions”, unless you mean the “instruction” to generate mRNAs that get spliced and regulated to produce protein base sequences. DNA is an anchor point in the dynamic dance of biology, one that gets copied into subsequent generations of cells, whther of progeny or somatic cells. That there are “instructions” applies to nothing physical. It applies to the ways we conceive of heredity, that’s all.
#2 and 3… Perhaps, for definitions of “think” so far removed from our normal use of the word as to render it virtually meaningless. I can speak of my computer toying with me, or my car not wanting to start on cold mornings, and I am anthropomorphizing them. Well, they also are; do they also think? Is there anything that is, that does not think? Is thought a property of matter? I think you can see that, while there are some philosophers who would argue idealistic monism, that in the vernacular, we use “thinking” to refer to a considerably smaller set of public and private behaviors. In our usual usage, I do not think that we can reasonably say that bacteria think. (BTW, in the history of psychology, Darwin’s writings opened up thinking in both directions–prior to Chuck, humans and animals were under no obligation to think similarly. After, common descent implied more options–A) thinking as humans know it is shared with other species (how many? remember planaria experiments?), B) human thinking is not really as we know it (it is an illusion–not that it is not there, but that it is not what it appears to be), as well as C) humans are the only species with our particular version of thought, although precursors may exist in other species. The situation JSW is bemoaning is, in my opinion, an acceptance of option A, brought about by unthinking acceptance of metaphorical speech as concrete. It is as if we took the word “sunrise” at face value, and explained it in terms consistent with the sun actually climbing in the earth’s sky.)
Many thanks for the link, John. In writing my post originally, I was torn between criticizing the anthropomorphizing, and pondering why it is that the vast majority of people find it difficult to not do so. Is it more of a deficit of language (that we need to borrow words that imply other things to describe the rest of biota), or is it a psychological quirk? Those two possibilities are related of course though…
I concede your point. Let’s extend the idea and see where it leads. It’s clear that if linguistic description is going to be left in the anthropological realm (which it must according to the original statements) then the properties of biological objects can be formally reduced to mathematics. Can we, and if so how, differentiate bacteria “thought” (thought being defined as a functional transformation), from human “thought?” To do so, at minimum, we require that the computational operation for each biological object (bacteria/human) is its own proper class, and that proper class is mathematically defined. Maybe an entire new taxonomy is needed; maybe based on a biological objects’ receptivity and capacity to affect a particular class of transformational operations? And if such a classification system can be established, what problem could it solve…?
John: That *is* what I meant when I used ‘instruction’ :). In computer science-y doublespeak, instructions are generally another word for ‘operation’, and operations (given a certain set of inputs, return a certain set of outputs) are the “basic building blocks” of large swaths of the theory behind the field. From what I can tell off that wonderful source, wikipedia, throughout the lifetime of a cell its DNA yields mRNA that produce proteins that specifically shape the function of the cell. Modifying a cell’s DNA changes what proteins get produced, and in turn, what other parts of that cell’s DNA get “read” or “written”. DNA, then, participates in a form of computation, and its intricate patterns necessarily hold information (the set of chemical operations necessary to yield amino acid xyz). Broadly speaking, information can be regarded as any relationship between a given set of atoms/basic symbols/attributes/whathaveyou in the system you are examining. The rings in tree bark still tell you how old it the tree is, and people being around to interpret that fact doesn’t change the underlying relationship. Wordplay is tricky! I wouldn’t call natural selection _intelligent_, but it certainly acts as a rational agent (which has a far more strict definition), etc. Maybe we’re just being imprecise?
hmmm.. i dont think that’s true. it’s just a pre-programmed process that is embedded on the bacteria for survival purposes.
I don’t usually comment here because science blogs are boring, nothing about hunting rabbits or sniffing out bitches in heat or useful stuff like that! However this cognition thingy is quite an interesting topic of discussion. You know there are dogs that argue that Thony C. (he’s my man-servant) can think because not only can he open tins of dog food but he even does so at the right time i.e. when I’m hungry. Now it only takes a moments thought to realise just how fallacious this argument is. It’s just conditioning! I’ve trained him well and when one watches him perform one might get the mistaken impression that there is actual thought behind his actions but as I have already said it’s just conditioning. I think you are quite right we should do everything possible to stamp out this sloppy caninomorphic argumentation. Every dog knows that only dogs are capable of intelligent thought and humans are just creatures of instinct. Now how about a discussion on rabbit hunting? Yours Sascha.
Thank you for another interesting post! If Have Question though: If I read your post correctly, you don’t like antropomorphizing, mostly because it introduces a notion of purpose in fields that are better of without it. Now, what definition of cognition do you use that implies goal-orientedness?
John It may be that the human concept of ‘concept’, ‘think’, ‘instruction’, etc. is simply another instance of the illusion of personal duality. Because our brains have no real way of being aware of themselves, the processes within our brains seem to be distinct from the physical/chemical processes occurring within our bodies. This apparent duality then leads us to ‘think’ that our ‘thought’ processes are somehow different from the chemical/physical/biological processes that we observe in nature. I conclude therefore that there is no reason to suppose that the processes that we call ‘thought’, ‘intelligence’, etc. are any different than other apparent patterns in natural processes of any kind.
When do metaphors assist in communicating Science? When do metaphors hinder in communicating Science? When does narrative assist in communicating Science? When does narrative hinder in communicating Science? I think that these questions have nuanced answers, depending on what kind of Science, what kind of audience, and what is the goal of the communication. I’d point out that bacterial cells and viruses outnumber human cells in and on the human being. If human cells were transparent and bacteria were opaque, human would look roughly the way they already do. The human/bacteria/virus symbiote, loosely called a human being, does indeed think. But, of course, that is not the meaning of the purple prose that you quite reasonably critique. On the higher mathematics blog n-Category Cafe there was a thread which I parapharse as: When do metaphors assist in researching and communicating Mathematics? When do metaphors hinder in communicating Mathematics? When does narrative assist in communicating Mathematics? When does narrative hinder in communicating Mathematics? Do you, John S. Wilkins, see a meaningful difference in the role of metaphor and narrative in Biology, by contrast with Chemistry, Physics, Astronomy, Anthropology, Psychology, Sociology, Mathematics? Between refereed science journals, conference proceedings, textbooks, press releases, science fiction, bestselling popular science books, web sites, blogs? Interesting bundle of issues that you raise. “Bundle” and “raise” both being kinaesthetic metaphors, which, by the way, were central to the creative process of Albert Einstein and Richard Feynman.
When do metaphors assist in communicating Science? When do metaphors hinder in communicating Science? When does narrative assist in communicating Science? When does narrative hinder in communicating Science? I think that these questions have nuanced answers, depending on what kind of Science, what kind of audience, and what is the goal of the communication. I’d point out that bacterial cells and viruses outnumber human cells in and on the human being. If human cells were transparent and bacteria were opaque, human would look roughly the way they already do. The human/bacteria/virus symbiote, loosely called a human being, does indeed think. But, of course, that is not the meaning of the purple prose that you quite reasonably critique. On the higher mathematics blog n-Category Cafe there was a thread which I parapharse as: When do metaphors assist in researching and communicating Mathematics? When do metaphors hinder in communicating Mathematics? When does narrative assist in communicating Mathematics? When does narrative hinder in communicating Mathematics? Do you, John S. Wilkins, see a meaningful difference in the role of metaphor and narrative in Biology, by contrast with Chemistry, Physics, Astronomy, Anthropology, Psychology, Sociology, Mathematics? Between refereed science journals, conference proceedings, textbooks, press releases, science fiction, bestselling popular science books, web sites, blogs? Interesting bundle of issues that you raise. “Bundle” and “raise” both being kinaesthetic metaphors, which, by the way, were central to the creative process of Albert Einstein and Richard Feynman.
Phill MV wrote I wouldn’t call natural selection _intelligent_, but it certainly acts as a rational agent (which has a far more strict definition), etc. A commenter named Febble (a Ph.D. neuroscientist, in fact) was banned from Bill Dembski’s Uncommon Descent for pointing out that natural selection precisely meets Dembski’s definition of “intelligent.” 🙂
Wow, I just picked up Debra Niehoff’s The Language of Life and found it the most insanely stretched collection of biological metaphors that I’ve ever tried to wade through. I completely agree with this post. It only makes concepts harder to understand on their own terms when the metaphors start to replace reality. Sometimes I think we should just make up new words instead of using metaphors to explain processes, but that would be perhaps going too far.
Wow, I just picked up Debra Niehoff’s The Language of Life and found it the most insanely stretched collection of biological metaphors that I’ve ever tried to wade through. I completely agree with this post. It only makes concepts harder to understand on their own terms when the metaphors start to replace reality. Sometimes I think we should just make up new words instead of using metaphors to explain processes, but that would be perhaps going too far.
Two kinds of anthropomorphising: 1) “Gee, these ants act in ways that are formally similar to my actions. Therefore, they are like me.” Translation: the world must be like me. 2) “Gee, these ants are nothing like me. Therefore, even though they are acting formally similar to my actions, they must not be doing that thing I am doing.” Translation: the world can’t be like me because I’m special. In both cases, the logical fallacy comes from assuming that the human is the measure of the world. In other words, anthropomorphism assumes either (maybe both) 1) cognition in other critters must be like humans, or 2) cognition in other critters must be unlike humans because cognition is always already human cognition. Same mistake, different disposition of the person making it. It’s too bad that so many scientists are die-hard humanists in this manner. Aren’t biofilms amazing regardless of their similarity to always-already-human cognition? Isn’t actual human cognition amazing regardless of its dissimilarity to biofilms? Before you can encounter another critter in such abstractions as ‘cognition,’ it should be encountered in the material specificities of what it actually does. IMHO, that is.
John: Humans reason by analogy, and are condemned to use language, symbols, and concepts that are mediated by human experience and biology. In this sense, all of our mutterings necessarily have anthropomorphic underpinnings. That doesn’t bug me, so long as it’s self-conscious or admitted to after the fact. If someone thinks “human thought” is the analogy that best fits for communicating an understanding of how bacterial films adjust to their environment, that’s OK by me, so long as they can explain the functional equivalence. If you think there is a better analogy, or that this one doesn’t hold water, let’s hear why. Of course, the claim that human neural networks and bacterial films are doing the same thing, is just silly. What may NOT be silly, is a claim that there may be important functionally isomorphic elements shared by both kinds of networks. Functional isomorphism does not depend on an equivalence of form. Though very different, analog and digital computers can both do functionally equivalent things.
And just to complicate things, teleology != thought. Even without thought, I’d offer that evolution can in fact create teleology de novo — not overarching purpose, but local goals accreted by successive specializations and other “commitments” to a given way of doing things. My basic example is that once you’ve specialized for say, the niche of gazelle, “speed is good” in no longer an option, it’s a fact of life. Even if the environment changes around the species, that “speed is good” theme will have woven its way into everything from the digestive system to mating rituals. With regard to the “thought issue” — what we do is a particular kind of information processing, honed by a long series of macroscopic, discrete, independently reproducing animals, and specifically by a line which selected heavily for social interactions. Something like an anthill, or a biofilm, may well be processing just as much information on their respective scales… but they’re not processing it the same way, at all! (Neither, of course, is a computer.)
I’ve been away a couple of days, so I’ll content myself with a general response. Analogies are useful for communication, education and advocacy. They are pretty well useless for actual science. An analogy adds only what you put into it, and licenses inferences only when it ceases to be an analogy and is backed up by a hard empirical study, or by a formal model or theory. Analogies are based on perceptions of similarities between a domain that is (it is supposed) known well and another domain that is not well understood. If, by lucky chance, the analogy happens to be isomorphic with the novel phenomena, fantastic (but you still need that empirical data and formalisation), but most of the time it won’t be, a priori. In this domain, as in biology in general, we are well past the time when we need to use analogies to understand phenomena. The reason why biology beats philosophy hollow for neologisms is that different words, without the prior connotations, are needed precisely because the biological phenomena are not much like our naive conceptions. And this is true even in psychology. So I think that we should not try to make overblown claims based on rough and ready subjectivism in our analogies. At least in biology. Maybe it works in physics, I don’t know.