7b. Cauchoix, M. & Chaine, A. S. (2016). How can we study the evolution of animal minds?
Reading: Cauchoix, M., & Chaine, A. S. (2016). How can we study the evolution of animal minds? Frontiers in Psychology, 7, 358.
Instructions for commenting: Quote the passage on which you are commenting (use italics, indent). Comments can also be on the comments of others. Make sure you first edit your comment in another text processor, because if you do it directly in the blogger window you may lose it and have to write it all over again.
7.b. I found this paper really interesting because it connects animal cognition to evolution in a clear way. The authors argue that mental traits, just like physical ones, can evolve through natural selection, which makes a lot of sense. I liked how they used real examples, like food-storing birds and brood parasites, to show how thinking abilities affect survival. But while the “fitness approach” sounds promising, it seems extremely hard to apply in real life since measuring cognition and fitness in wild populations isn’t easy (practical side of this paper feels a bit idealistic).
ReplyDeleteAnthony, comparative cognition and the evolution of cognition have become increasingly important parts of cognitive science, especially the evolution of learning, social communication and how they led to language in our species (only). What is “Baldwinian Evolution”?
DeleteAs mentioned in a previous discussion, a baldwinian evolution refers to the evolving of a specific trait or ability that is not initially inherited but rather learned and honed over generations, providing the species with an adaptive edge that feels like second nature and is ultimately encoded in the genome. With regards to language, I think of the theory of universal grammar, proposed by Chomsky, which suggest an innate language module that provide a flexible structure allowing the development of any given language in babies. It's thought of as a language acquisition device, not to be confused with language learning per se (vocabulary pertaining to x language). If we take a look at our distant ancestors, before verbal language as we know it was used, I am inclined to say that eloquence (as primitive as it was) might have provided some with an adaptive edge (easing communication, fostering inter-group harmony, strategically managing predators, etc.) leading to it being encoded in the genome. This would allow for the development of a now innate language acquisition capacity among humans, as Chomsky's theory would suggest, but I'm not sure Chomsky would like my origins theory on his theory...
DeleteFrank, ask GPT again what Baldwinian Evolution is. The capacity to learn is adaptive, and the genes that code for that capacity are hence inherited. But what you learn (or imitate) is not itself inherited, it is learned. What can evolve is the genetic capacity to learn certain kinds of things faster or more easily. And that’s what some call Baldwinian Evolution.
DeleteHow can there be “eloquence” before language? (You have to think more critically.)
We’ll discuss language evolution and Chomsky’s Universal Grammar in Weeks 8 and 9.
I was intrigued to see what an LLM would answer to the question "what is Baldwinian Evolution", especially to practice for the final exam. I asked Gemini and I am having a hard time finding any corrections to make. It said that this theory "suggests that an organism's learned behaviors or characteristics acquired during its lifetime can influence the direction of its genetic evolution without directly inheriting those learned traits", explained the steps, and used the example of birds being predisposed to learn certain types of songs. To me, Gemini correctly answered by making the important distinction that the things are still learned rather than inherited, and that what is inherited is the capacity to learn more easily and effectively. Am I wrong?
DeleteI think this paper shows that learning and thinking can be part of evolution too, not only body traits. Like Anthony said, animals that can remember things or solve problems can survive better. But I also understand what the prof said, that it is not what animals learn that changes in evolution, but their ability to learn. I think Baldwinian evolution means that being good at learning can become something in the genes, even if the thing learned is not passed on, so that makes me agree with Anne-Sophie that it is important to know the difference between learning something and being born with the skill to learn it. The bird example makes sense because the birds do not get the song from birth, but they are born ready to learn that kind of song. For me, the most interesting thing is how intelligence looks different in every species. It is not just humans that use learning to survive. Many animals do too, in their own way. I think this shows that being able to learn and change fast is one of the most important parts of evolution.
Delete***EVERYBODY PLEASE NOTE: I REDUCED THE MINIMUM NUMBER OF SKYWRITINGS. BUT THE READINGS ARE **ALL** RELEVANT TO AN OVERALL UNDERSTANDING OF THE COURSE. SO, EVEN IF YOU DO NOT DO A SKYWRITING ON ALL OF THEM, AT LEAST FEED EACH READING YOU DO NOT READ TO CHATGPT AND ASK IT FOR A SUMMARY, SO YOU KNOW WHAT THE READING SAID — OTHERWISE YOU WILL NOT HAVE A COMPLETE GRASP OF THE COURSE TO INTEGRATE AND INTERCONNECT FOR THE FINAL EXAM.***
DeleteAnne-Sophie, exactly right. Baldwinian evolution is still genetic, hence Darwinian, but it refers to when genes have made learning something that is especially important faster and easier to learn.
Rena, you are right too, and about how the ability to learn is an especially important adaptive trait. Evolution is "lazy": so it does not code genes for something the organism can learn. But when something learnable, Baldwinian evolution selects for genes that give its learning a head-start in speed, ease, and motivation to learn. That is the case not just in songbird learning if their species' calls, but in human language learning (and even duckling imprinting). How?
Instructor By going through the skywritings, I am noticed that you used "motivation to learn" in many instances when referring to Baldwinian evolution. While I understand that Baldwinian evolution is the process by which a genetically based capacity to learn (the ability to more easily useful capacities, e.g. language) provides an adaptive advantage, so that genes favoring those learning tendencies become more common over generations. I am having an issue with your use of the word "motivation". How to you go from genes that "give its learning a head-start in speed" (which I understand), to motivation to learn? Are you saying Baldwinian evolution "pushes" individuals towards that learning capacity by facilitating the learning curve?
DeleteThis paper argues that animal minds should be studied using the same evolutionary approach we use for any other biological trait. The authors explain that evidence shows animal cognition meets Darwin’s 3 necessary conditions for evolution by natural selection (variation between individuals, is heritable, and influences fitness). By comparing indicators of cognition across different species and measuring how variations in cognitive ability affect fitness within a single species, we can better understand and measure how their minds evolves. I found it interesting that this paper focuses on animal cognition, since the evolutionary approach in psychology was first introduced in the context of humans in 7a, and much of the course is centered on human minds. I wonder how these ideas could be applied to human cognition, and what differences or limitations might arise when extending this evolutionary framework to our own brains.
ReplyDeleteAnnabelle, yes, the focus of this cognitive science course is mostly on human cognition, but nonhuman animal cognition already made an appearance in Week 4, on mirror neurons and mirror capacities (what are they?). And the precursors of supervised learning in AI (what’s that?) are in reinforcement learning in animals. And, as we get to the "Hard Problem" in Week 10, we naturally reach the question of animal sentience in Week 11,
DeleteThis paper argues that animal minds should be studied using the same evolutionary framework we apply to other biological traits, since cognition meets Darwin’s three conditions for natural selection (variation, heritability, and fitness). This connects to our earlier discussion in Week 4 on mirror neurons and mirror capacities, which refer to the neural systems that activate both when performing an action and when observing another do it. These mirror capacities allow animals, and humans, to understand and even predict others’ behaviour, which likely offered strong adaptive value in social environments. The link to reinforcement learning is also clear, animals learn through feedback, by associating actions with rewards or punishments, which serves as a biological precursor to supervised learning in AI, where models are “trained” using labeled examples and corrective feedback. Both systems adjust behaviour or output based on experience.
DeleteAyla good synthesis, except reinforcement learning is biologically plausible but "labelled input" is not: Why not?
DeleteI agree that this is a great synthesis! Answering this question posed about why labeled input is not biologically plausible brings us back to what reinforcement learning is compared to supervised learning. Reinforcement learning, whether in animals or in AI, occurs through trial and error. The system (animal or model, etc.) makes an action, gets feedback (reward or punishment), and then adjusts its behavior to get a better outcome in the future. However, in supervised learning, the system adjusts its behavior by way of labeled examples instead of trial and error—for example, being told directly that "this is a cat." Animals don't get explicit labels in nature because they aren't told what's right or wrong explicitly in words, but infer them through good or bad consequences. Therefore, reinforcement learning is biologically plausible because it shows how animals learn from outcomes, but labeled input (used in supervised learning) is not biologically plausible because nature doesn’t provide labels for every action or stimulus encountered.
DeleteRachel H, that’s right (except machines are not “told”):
DeleteAll input-output learning with error-correction is “supervised”. It’s just that, in reinforcement learning, (1) input is followed by (2) output, followed by (3) +/- feedback (“right/wrong”) -- followed by (4) “error backpropagation,” which updates connection weights to make an error less probable and a correct output more probable.
In what has come to be called “supervised learning,” the difference is that each input comes "pre-labelled" with the correct output, and then backpropagation updates the connection weights to make an error less probable and a correct output more probable. This can only be done in artificial machines, in parallel batches, updates jointly.
Reinforcement Learning (RL): The machine (human or artificial) acts and gets feedback — success or failure. Backpropagation (BP) [ask GPT what that is] adjusts the machine's internal connections to do better next time.
Supervised Learning (SL): The machine (human or artificial) is given pre-labelled examples of right and wrong. BP tunes the network to match the labels. This is efficient for industrial machines, done in parallel batches, but it is unrealistic biologically.
In the relevant sense, both RL and SL are "supervised."
There is also Unsupervised Learning (UL): what is that?
And there is also something (misleadingly) called "Self-Supervised Learning" ("SSL"). What is that? And what is misleading about calling it SSL? It's really just another form of SL. (Hint: an example is the next-token training of LLMs on the "Big Gulp" database.)
What I find especially interesting about this discussion is the compare and contrast between animal learning and machine learning. In the reading, Cauchoix and Chaine (2016) note that when we study animal cognition from an evolutionary perspective, we are often trying to understand how aspects of cognition such as problem solving, memory, and learning from experience are used to enhance fitness. Reinforcement learning seems to embody this process well, because animals (or humans) do not get "labelled" examples, they learn from their consequences such as what events help them to survive or reproduce.
DeleteBy contrast, supervised learning feels like a fabricated shortcut, it's efficient for machines but biologically inaccurate because nature does not provide labeled examples. The unsupervised learning aspect might be the part that animals find out and detect meaning or regularity/patterns in their experience without feedback learning, like seeing textures or rhythms in their surroundings. Perhaps what is a little misleading about "self-supervised learning" is it is still a bounded way of "being structured", the system creates its own label but it is not learning something from scratch. It is interesting to think about all of these as being examples of different learning strategies that may have evolved in nature.
Professor, reinforcement learning is biologically plausible because, as the paper explains, animals adjust their behaviour through experience, linking actions with outcomes that affect fitness. They learn through feedback signals like success in finding food or avoiding predators, which mirror the trial-and-error processes shaped by natural selection. However, labelled input assumes an external source explicitly tells the learner what the correct output should be, which doesn’t exist in nature. As the reading notes, cognition evolves through feedback from environmental and social consequences, not through pre-defined answers. In other words, animals learn adaptively from reinforcement, not instruction.
DeleteI am not sure if I am correct, but I think self-supervised learning is a type of supervised learning that creates labels on its own from data that is unlabeled. LLMS have training based on huge amounts of text and the model predicts the next token based on the Big Gulp. So, the next word already exists in the data and becomes the label, reducing error in prediction through backpropagation. The name is misleading because it acts as if this is a new autonomous form of learning when it’s just error-corrected training based on labels through automatic text generation.
DeleteUnsupervised learning may be a type of learning where the machine is not given any labeled data or any form of feedback. It may just find patterns from the input it receives and organize data based on that. It’s close to how we humans detect common patterns in the environment without instructions (e.g. human heads are round, apples are red).
Cauchoix and Chaine (2016) argue that animal cognition can be subject to natural selection (a mechanism within evolutionary biology), like other traits, by providing evidence that cognitive traits vary among individuals, are heritable and affect fitness. They further discuss that researchers often use between-species methods to understand how broad patterns in cognition are linked to ecology in animals. However, to get closer to causality, they suggest complementing between-group comparisons with within-species studies that look at contemporary selection. These studies adopt a “fitness approach” by attempting to link variation in cognitive functions within a species in the wild to ecology-relevant behaviours, and in turn, to changes in survival/reproduction, while considering the socio-ecological context they live. That said, what methods do you guys believe would best help inform how humans’ cognition is shaped by evolution today?
ReplyDeleteFor humans, I think the “fitness approach” becomes tricky since survival and reproduction aren’t our only measures of success anymore. But maybe we can still study how cognitive traits like cooperation, risk taking, or social learning influence outcomes in modern environments.. things like career achievement and innovation etc. It’d be cool to see whether those patterns reflect adaptive remnants of ancestral selection pressures or entirely new forms of selection shaped by culture and technology.
DeleteGabriel I think that today the effects of our (genetically) evolved cognitive capacities, especially language, are much more pronounced and rapid (and radical) than the effects of our environment on our genes.
DeleteShireen, reply is similar to the reply to Gabriel; be careful to distinguish genes and "memes". Memes need hundreds of thousands of generations to affect genes, but they can affect memes in as few as one generation. Why?
Because ideas can spread and change quickly through learning and communication, while genes can only change through reproduction over many generations. You’re pointing out that for those same ideas to shape our biology--to actually alter our genes--it would take many generations of selection.
Delete“We describe how cognition can be subject to natural selection like any other biological trait and how this evolutionary approach can be used to understand the evolution of animal cognition.” I found this idea really interestin, that cognition itself can evolve through the same principles as physical traits. It reframes intelligence not as something mysterious or exceptional, but as a biological adaptation shaped by variation, heritability, and fitness consequences. But how we can distinguish between traits that are truly cognitive adaptations versus those that are flexible byproducts of other evolved systems. Like when an animal learns a novel behavior, are we seeing selection for cognition itself or just selection for behavioral flexibility in general?
ReplyDeleteShireen, cognitive capacities (like learning or language) can evolve genetically if they have a genetic basis. But what an organism learns or says by means of its evolved, genetic capacities is not necessarily itself genetic -- although the disposition or motivation to learn to do it may be enhanced genetically by "Baldwinian Evolution," (What is that? Ask GOT and come back and tell us.)
DeleteI also feel like it is compelling to consider intelligence as something that is shaped by evolution instead of just an abstract mystery. I think what is challenging is trying to determine what is actually a true cognitive adaptation and what is just a ‘byproduct’ of other traits. Like when animals use a new foraging strategy, is evolution selecting for its ability to think in that specific way, or is it selecting for a more general capacity to adjust behavior based on its experience. It also makes me wonder about humans: how much of what we can know is actually a direct target of selection, versus an emergent property of systems that evolved for other reasons (like social coordination or memory survival tasks). To me, it highlights just how intertwined behavior, cognition, but also ecology actually are and how careful we have to be when claiming something is an adaptation of the mind itself. But, this also makes me think of a bigger question: if some of our cognitive abilities evolved for survival instead of well-being, should we trust our instincts and reasoning as much as we do, or, question them in modern contexts where the challenges are mostly abstract and social rather than physical and based on survival??
DeleteLauren, good reflections. It's all biology, of course, and all a product (directly or indirectly) of evolution. For some behaviours it's easy to determine whether they are inborn or learned. For others it's not so easy, though that doesn't mean that there too there aren't some that are learned and some that are inborn. But you're right that behavior, cognition, and ecology are intertwined. (I'm not sure what you mean by the difference between what evolved for survival and for "wellbeing"...)
DeleteLauren, I really like your point about how our evolved cognition might not always align with modern well-being, but I also think it’s interesting how those same mechanisms continue to serve adaptive purposes, just in new forms. Cauchoix and Chaine describe intelligence as a flexible, problem-solving system shaped to handle unpredictability, and I think that flexibility still applies today. Even though our environment has shifted from physical to social and technological threats, we’re still using the same cognitive architecture to anticipate danger, form alliances, and manage uncertainty. Our intelligence may not have evolved for abstract reasoning, but it allows us to adapt to an increasingly complex world. In that sense, I think that we can still trust our instincts and reasoning that evolved for survival, it’s just that the “predators” we face now are the modern issues we face today.
DeleteWell-being here seems to mean something closer to mental health or a sense of flourishing and how our cognition supports not just staying alive, but living meaningfully and sustainably in complex societies. The tension she’s describing is, I think, between traits that kept our ancestors alive and those that help us feel fulfilled or stable today.
This reading (and the 7a reading) both argue that to get a more clear understanding of how animal minds evolved, connecting how minds work (proximate) with why behavior evolved (ultimate). Evolutionary psychology is the field that focuses on this by trying to explain how mental adaptations are shaped by natural selection. This field uses fitness approaches, how cognitive traits evolve by looking at how they affect an animal’s survival and reproduction, and comparative approaches, how traits evolved by comparing living species while considering their shared ancestry from a phylogeny, to study animal behavior in a lab or in the wild. Overall this could lead to new discoveries about the interaction between genes, environment, and behavior, showing that cognitive traits are subject to selection.
ReplyDeleteSierra, how has the evolution of the (genetic) capacity for learning and language made it harder to determine whether a behavior is evolved or learned?
DeleteThe evolution of the capacity for learning and language has made it harder to determine whether a behavior is evolved or learned because the line between what is innate and what is learned is blurry. In fact, as the authors explained, cognitive abilities such as perception, attention, decision-making, executive functions, learning, and memory are products of evolution. However, these traits are subject to change according to one's environment/experience. This can be seen in people's different abilities to speak different languages. For instance, the majority of people are born with the capacity to learn language however, not everyone will speak the same language. While some will have the ability to articulate different sounds, it will be significantly harder for others. This is all shaped by experience and learning. Thus, I think our behavior is usually a mix of both evolution and learning.
DeleteNada I appreciate your comment on determining whether a behavior is evolved or learned. I agree that Baldwinian evolution and the "increased capacity" to learn specific abilities (e.g. language), which is by itself evolved but enables for learned behavior, it is harder to determine between learned and evolved. However, I am not sure I fully grasp your language example, I will try to deconstruct it : (1) I agree that humans are born with the capacity to learn language, (2) however, from my understanding, every verbal human has the capacity to learn to articulate different sounds. The moment it becomes significantly harder for others is once we are passed the critical language period. (3) I guess by writing this comment I understand more your point as you are saying that the critical language period refers to Baldwinian evolution, which are evolved genes that allow for faster learning, i.e. enhancing learned behavior.
Delete"We describe how cognition can be subject to natural selection like any other biological trait…”
ReplyDeleteThis argument is convincing, but it might simplify what “cognition” really means. Unlike physical traits such as beak length or brain size, cognition depends on the context and often comes from interactions between the brain, body, and environment. Treating it as a single, measurable trait assumes the mind works like a machine. Wouldn’t it make more sense to study whole “cognitive ecologies” instead of isolated mental traits to understand how minds evolve?
Randala, what is a machine? Biology (including cognitive science) is trying to reverse-engineer the causal mechanism that produces both the structures and the functions of organisms. Isn't that what is meant by machine. This is not do deny that organisms are sentient machines -- and that reverse-engineering the causal mechanism that produces feeling is a "Hard Problem"...
DeleteFigure 3 caught my attention. It shows a framework that assumes we can clearly separate cognitive abilities (eg attention, memory), behaviour (parental care, defense) and fitness (number of babies?) into separate categories. But does this model not oversimplify cognitive functions and behaviour as separate entities when, in reality, they often are intertwined? For example, the paper says that memory and attention are often coupled together neurally; so can we really assume they evolved independent of one another? Furthermore, some behaviours, such as parental care, perhaps don't come only from flexibility + memory + attention, but as a result of some interaction of these processes.
ReplyDeleteI agree, Elle. Figure 3 does make cognition look too modular. But I think Cauchoix & Chaine separation is mainly an analytic shortcut rather than a literal map of how the mind works. Their goal seems to be knowing the strongest selection links in the chain which could be cognitive, behavioral, or ecological. In practice, attention, memory, and flexibility almost always engaged at the same time, but conceptually separating them aids researchers in tracing the effects of natural selection and fitness. I read the model as scaffolding: simplified enough to test, yet meant to be rebuilt once those interactions are empirically measured.
DeleteI agree, figure 3 does risk looking as though it divides cognition, behaviour and fitness into neat, isolated “boxes”. However, as the paper points out, the framework is meant more conceptually. The authors make an emphasis on how cognition, behaviour and fitness are deeply interdependent, represented by the bidirectional arrows since they influence each other.
DeleteBehaviours like parental care and defense aren’t produced by a single trait such as memory or attention but emerge from interacting cognitive processes shaped by these ecological pressures. The model is meant to help one understand how cognition can be linked back to fitness, but also clarifies that real biological systems are much more integrated and dynamic (difficult to display in a simple diagram). It’s more a scaffold for connecting the differing levels of analysis being cognition, behaviour, and fitness showing how they co-evolve and have mutual feedback.
The article provides a foundation for future inquiry into the evolutionary basis of cognition. It explains that in order to understand how our mind works, we need to connect it to why such cognitive traits exist. Chauchoix and Chaine, using recent studies, highlighted how linking “proximate causes” (neural mechanisms) to “ultimate causes” (evolutionary pressures) helps explain cognitive mechanisms, the easy problem to cognition. However, the hard problem, why, or how consciousness arises from those mechanisms, is still left unanswered even through research into evolutionary cognition. Unless, through this inquiry we discover that while natural selection shapes information processing it also shapes how it feels like to process that information. Perhaps feelings themselves are adaptive signals used to tune cognition towards survival.
ReplyDeleteI agree and I think a core element that the authors return to in pursuing this line of research, is the need for direct measures of cognitions. It seems like cognitive science in general has hit this wall in research: we are lacking new tools to probe/measure/understand brain functions. And the results struggle to offer definitive evidence for human cognition because strong causal experiments are relinquished to animal studies due to ethical concerns. Especially in the case of genetics, which (for many good reasons) we have decided that messing with the human genome is unethical for the sake of scientific exploration. I am hopeful that merging evolutionary biology and cognitive science will give rise to some undoubtedly interesting research, but I’m not convinced that given the current tools at hand, it will revolutionize our concept of cognition. I also humbly acknowledge that I am merely a skeptical undergraduate student and am in no way qualified to offer a critique on the entirety of cognitive science, so this is merely a pondering.
DeleteI found very interesting the example of aviary brood parasites, as it illustrated a form of competitive cognitive evolution. Brood parasite’s attitude of abandoning eggs in foreign nests have led to the development of larger hippocampus in the nest-searching sex of parasitic species, allowing for better spatial navigation and recognition of potentially successful host-nests. This then led to the development of increasingly refined neural structures and cognitive mechanisms in non-parasitic species to allow for the implementation of a set of strategies to detect invasive chicks. I would be curious to see if anything similar (likely to a much smaller scale) has occurred in the case of humans and species that have evolved next to them (e.g. dogs).
ReplyDeleteCauchoix and Chaine discuss how cognitive traits in animals evolve through natural selection, just like physical traits do, and they need to show variation, heritability, and fitness consequences. They argue that researchers should measure how selection actually works on cognition in wild populations instead of just studying mechanisms in labs.
ReplyDeleteThis made me think about the Turing Test we discussed in class. Professor Harnad said the Turing Test is about designing a mechanism that can do everything humans can do, but Cauchoix and Chaine show that cognitive abilities evolved to solve specific survival problems in particular environments. So maybe a model passing the Turing Test wouldn't just need human-level performance, but would need cognitive capacities that evolved under similar selection pressures that humans faced.
I agree, shit happens and here we are. Evolution undeniably played a role in “shaping” the human brain’s cognitive abilities of today (to what extent?). It involved learning how to categorize stuff and “do the right thing with the right kinda thing” for better survival rates. Bringing back the mushroom island, learning the different features of mushroom A, B and C required direct sensorimotor interaction with the environment. The most fit were the one who could categorize and then combine the categories via language (by grounding symbols). Over the time, it was not necessarily a matter of environmental context, but rather a question of “having it in the genes” or “being able to learn how to”. Today, categorization allows us to tell stories, solve mathematical problems and differentiate a ripe apple from an unripe apple. This is stuff humans can do. What does it mean for creating machines and systems? Would they need these colours sensors that are now inborn in humans? Could they have a capacity to learn all sorts of languages? How can we speed up evolution if they do not need to learn in context of survival and reproduction? Would we need to speed it? One thing is sure, they would need direct sensorimotor grounding and a physical body.
DeleteI found the Cauchoix and Chaine (2016) article an interesting complement to the Lewis et al.(2017) reading on evolutionary psychology. While Lewis et al. focused on how psychological mechanisms may have evolved to solve adaptive problems, Cauchoix and Chaine approached cognition itself as an evolving trait. Both perspectives share an interest in linking mind and evolution, but they differ in scope: Lewis et al. work primarily with human behavior, while Cauchoix and Chaine emphasize cross-species comparisons and ecological contexts. I appreciate how Cauchoix and Chaine propose studying cognition through heritable variation and fitness, yet I wonder how realistic this approach is in practice, given the difficulty of measuring cognition objectively across individuals or species. Can we truly identify selection acting on cognition without reducing it to performance in specific tasks? This tension highlights both the promise and limits of applying evolutionary frameworks to the study of minds.
ReplyDeleteOrganisms that have the capacity to learn better (faster, with fewer repeated mistakes) would survive longer than those who learn poorer and go on to reproduce more (Baldwinian evolution). This does seem like natural selection/Darwinism, since the fittest are the ones surviving, but it’s not because of an obvious physical superiority like running faster or hiding better or having bigger teeth. Instead, having better cognitive capacities specifically related to learning means having a trait of better adaptability. But how are cognitive traits passed down genetically? And, in practice, how would researchers separate behaviour, cognition and fitness?
ReplyDelete"Indeed, much of behavioral or evolutionary ecology theory is based on strategic decision-making… there is little understanding of how information is processed and how cognitive abilities enhance or constrain decisions based on the available information"
ReplyDeleteI really like this passage because it shows how many studies focus only on what animals do instead of how they think when doing it. It’s like judging someone’s choices without knowing how they decided. I think that understanding the thought process behind behavior is crucial because two animals might act the same way for completely different reasons (maybe one relies on memory, while the other relies on instinct). If we ignore those differences, we miss a big part of evolution. The way thinking itself evolves and adapts to new challenges is what truly shows the mind’s role in evolution. I wonder how researchers can study and uncover this without disrupting natural environments.
You make a great point, Jad! I feel like this all goes back to the big question of “how” and “why” thinking organisms can do what they do. Cauchoix and Chaine (2016) bring forth several cognitive processes such as spatial memory, learning, and attention networks which have yet to be explored across species. Similarly to brain studies in humans, most work on cognitive performance is correlational, straying away from causal explanations of behaviour that may allow us to reverse-engineer the animal brain. But as you point out, it is important to distinguish between different types of cognitive processes, as each may serve diverse adaptive functions that contribute to an organism’s fitness. For example, one’s ability to problem-solve may have evolved to help locate food or avoid predators—revealing not only what animals do but why those behaviours persist from an evolutionary perspective. This is why the authors emphasize using a fitness-based approach, which focuses on underlying variations in cognitive abilities (and their influence on survival and reproductive success) to better grasp the evolution of cognition.
DeleteI find this incredibly interesting as a blend of multiple disciplines to explain how animals think. However, I wonder if this is able to avoid the pitfalls of brain studies. Just as with brain activity, there are myriad ways that the environment can change and affect fitness. Thus, how can we ever be sure that what we are seeing is actually affecting how animals think in the way that we believe it does? Again, drawing parallels with what Fodor said regarding brain studies, it does seem like quite a few of these (but not all) can be summed up as "the environment affects how animals think" or "these things made it so that this behavior occurs." So, while this approach is very good at explaining the why, I struggle to understand if it could even explain the how. What are the internal processes that lead to this behavior? Essentially, this approach to me seems like a good tool, but not the full answer to figuring out how to reverse engineer the animal mind
ReplyDeleteCauchoix and Chaine (2016) argue that the evolutionary study of animals is not complete with the study of behaviour only. They argue that the incorporation of cognitive capacities to evolutionary study of animals would allow for a better understanding of evolutionary traits. Thus, the authors show that, just like physical traits, cognitive capacities are also subject to natural selection and to adaptive problem solving. I find this approach highlights the complexity of living beings, by not just reducing behaviour on input, but by studying how animals process information and their decision-making process.
ReplyDeleteSome animals get better at surviving because they can learn and think. Cauchoix and Chaine (2017) explain that mental abilities, like remembering or solving problems, can change over time through evolution which helps animals find food, stay safe, and raise their babies. I think this is true because being able to think gives animals more choices. But it is not easy to study, since it is hard to know how much thinking really helps in nature. Every species learns in a different way, and scientists cannot always test that well. In the end, the paper shows that intelligence is not something random. It grows and changes like any other trait that helps a species live and survive.
ReplyDeleteThe reading suggests integrating proximate and ultimate perspectives in comparative cognition to understand how animal minds evolved. For kid-sib, proximate studies are HOW cognitive mechanisms work, while ultimate studies are the WHY that drives selection on these traits.
ReplyDeleteHowever, one important challenge to this integration involves methodology and ethics. I found the paper’s discussion of animal experimentation to be interesting. An issue with methodology is that animal experiments in laboratory settings introduce significant stress, a confound, which can impact the animal’s behaviour and cognition. This can raise doubts about the accuracy of such lab experiments.
As someone involved in animal research, I do find it ethically challenging. While a lot of clinical research relies on animal models, it can feel unjustifiable to introduce such stress in the rodents in non-clinical research when there are no tangible benefits.
Cauchoix and Chaine's paper talked about understanding evolution beyond visible behaviors emphasizing that cognition (how animals perceive, learn, and decide) also evolve under natural selection. By tying neuroscience with ecology they suggest that intelligence and decision making are adaptive tools shaped by environmental challenges, reframing evolution as not only a story of physical survival but also of mental innovation. They outline some methods such as comparing brain structures across species and linking problem solving abilities to fitness in natural environments. They also discuss the fitness approach, which connects cognitive performance to real survival and reproductive outcomes, though measuring cognition in wild populations still remains difficult given the fact that there can be many environmental and social factors that can interfere.
ReplyDeleteWhat struck me most about the paper is how confidently it treats cognitive mechanisms as traits that evolution can shape, even though we still have only a partial grasp of what those mechanisms actually are. We can measure behavior, and we can map broad brain regions, but the computations that transform perception into action remain largely unknown - especially in humans, the species we supposedly understand best.
ReplyDeleteThat mismatch made the authors’ proposal feel exciting but also ironic. They urge us to trace selection on attention, memory, and decision-making as if these were clearly defined traits, when in reality we haven’t reverse-engineered the algorithms behind any of them. We talk about “attention” as if it were one thing, but it’s likely a combination of interacting processes. We speak of “memory” as a trait when we don’t even know what the relevant unit of memory computation is.
And if cognition really does evolve by reshaping these hidden processes, then humans today are almost certainly undergoing cognitive selection in ways we can’t yet see. Our environments are shifting faster than any nervous system can comfortably adapt, but without understanding the underlying computational machinery, it's difficult to tell which cognitive capacities are being favored, compromised, or repurposed.
The paper left me with the idea that we’re trying to chart how selection acts on mechanisms that we’re still in the process of discovering. Is the big mystery here that we don’t know how cognition evolves, or that we don’t even know what’s evolving?
When the author mentioned high-impact, low-frequency adaptive problems driving certain traits, it made me think of The Moro Reflex in infants (the reflex seen in newborn babies wherein they throw up their hands in a grabbing motion when being dropped). Evolutionary scientists believe that this reflexive motion, which disappears after a few months of age, was a result of evolution (I would assume, Darwinian). I don’t believe that The Moro Reflex was driven by high-impact, low-frequency problems; in fact, I believe quite the opposite. It would make sense to genetically encode the fear of falling in humans who have not known the threat of a fall yet so that they don’t have to learn it. However, I’m wondering if its “opposite counterpart”, irrational phobias, could be linked to Darwinian evolution and the concept of (perceived) high-impact, low-frequency problems giving rise to certain psychological behaviours. Irrational phobias are different from rational fears in that, most of the time, they are rare in frequency of occurrence and are learned almost instantaneously (i.e., one encounter with the fear can result in a life-long severe phobia). Could it be that the ability to learn to avoid certain fears very rapidly (from even one encounter) be a Darwinian evolutionary trait?
ReplyDelete