Could people please stop reducing evolutionary phenomena to single, all-encompassing causes? Like, now?

Imagine you are a computer scientist and engineer, and you design a simulation that consists of a 15×15 square grid, with a ‘predator’ who follows some simple rules to seek out ‘prey’. You put most of your effort into designing the simulated ‘prey’ who uses visual detection, as well as some interesting uses of memory to simulate planning, to avoid the predator, and then you randomize the grid with various densities of black squares that block vision and white squares that allow line-of-sight to the predator. You work out some general principles for controlling the ‘prey’ simulation, for instance that less cluttered grids select for ‘prey’ that do more careful planning and have more complex rules for behavior.

I think that’s an interesting result, especially since they quantify everything. But is it enough to get a big splashy publication in Nature? To get noticed in the popular news media? Nah, probably not. It’s narrow niche research, but no one outside computer science is going to be impressed. So, how to spice it up?

I know! Claim that your simulation is significant evidence that explains the Cambrian explosion, the diversification of terrestrial vertebrates, and the evolution of human intelligence! That’s the ticket!

After explaining how their simulation works, the authors get down to explaining why their result is important. Apparently, it’s not because they’ve done anything interesting in computer science, so they need to borrow from biology to find a justification.

Parker has suggested that the origin of the Cambrian explosion lies in the atmosphere or oceans of the period gaining higher transparency to sunlight, triggering the evolution of the first image-forming eye and sparking a predator–prey evolutionary arms race that gave rise to the Cambrian’s profusion of animal forms. A second great change in transparency occurred with the emergence of fish on to land, which gave rise to a sensorium large enough to fit multiple futures. Our idealized model of spatial planning during predator–prey interactions suggests that there may be a link between the enlarged visual sensorium and habitat complexity of terrestrial animals and the evolution of neural circuits for dynamic planning.

They get even more hyperbolic when talking to reporters on CNN (see, it was a smart decision to add all that evolution stuff — it got the attention of the media).

The ways our ancestors adapted to live in patchy landscapes cluttered with obstacles “poured jet fuel” on the evolution of the brains of animals and early human ancestors, according to researchers at Northwestern University.

The combination of our enhanced eyesight and higher intelligence to survive in this complex land environment is “why we can go out for seafood, but seafood can’t go out for us,” said Malcolm MacIver, a professor of biomedical and mechanical engineering in Northwestern University’s McCormick School of Engineering.

Kind of a grand leap from a simulation on a checkerboard to an elaborate umbrella hypothesis that explains a complex and diverse evolutionary phenomenon as the the product of being able to see farther, isn’t it? I would have rejected this paper at the first sentence of the abstract, which tells me they’re not very knowledgeable about the biology they’re using as a prop.

It is uncontroversial that land animals have more elaborated cognitive abilities than their aquatic counterparts such as fish. Yet there is no apparent a-priori reason for this. A key cognitive faculty is planning. We show that in visually guided predator-prey interactions, planning provides a significant advantage, but only on land. During animal evolution, the water-to-land transition resulted in a massive increase in visual range. Simulations of behavior identify a specific type of terrestrial habitat, clustered open and closed areas (savanna-like), where the advantage of planning peaks. Our computational experiments demonstrate how this patchy terrestrial structure, in combination with enhanced visual range, can reveal and hide agents as a function of their movement and create a selective benefit for imagining, evaluating, and selecting among possible future scenarios—in short, for planning. The vertebrate invasion of land may have been an important step in their cognitive evolution.

It is “uncontroversial”? To whom? Which is “smarter”, a spider or an octopus? Why are you lumping the diversity of terrestrial animals into a neat tidy bin labeled “land animals” and making the assumption that they have more elaborate cognitive abilities than the wet creatures you’ve thrown into a bin labeled “aquatic animals”? If your hypothesis were sufficient to explain major evolutionary transitions, why is it that cetacean brains got larger as they adapted to an environment with reduced visual range? Why do you need a single a priori reason to explain the origin of complex phenomena with widely varying solutions? Why is it that the mantis shrimp, the animal with the most sophisticated visual sensorium, is neither particularly brainy nor terrestrial?

As you might guess from all of my questions, this is an immensely frustrating paper — not in its methods, or in the execution of the study, but in the overblown interpretations the authors have been made. It desperately need input from visual neuroscientists and evolutionary biologists, who might have put a damper on the nonsense they’ve shoveled into the work…but then it wouldn’t have made it to CNN, now would it? We could probably sell tickets to a battle of the umbrella hypotheses, Long Range Eyeballs vs. The Aquatic Ape. Epistemic humility just never sells.

(Oh, man, I keep coming back to exceptions they ignore. Aquatic and terrestrial environments are diverse — there are patchy aquatic environments and cluttered terrestrial ones. What about the issue of scale? The jumping spiders on my house live in a savannah-like environment, with wide open areas (the planks of my siding) of long-range visual opportunities, sprinkled with joins that are good hiding places. Why haven’t they invented calculus yet? If you tell me it’s because they’re tiny, then you’ve just admitted that the visual hypothesis is insufficient.)


  1. steve1 says

    That is an awesome picture of a mantis shrimp. Unbelievably some aquaristst don’t want them in their tanks.

  2. sarah00 says

    Oh god, I’m getting flashbacks to a failed PhD attempt that was derailed by supervisors who thought they had an umbrella hypothesis (a term I’ve not heard before but fits perfectly) and refused to listen to anyone, especially me, who told them that it wasn’t going to work. I had physicists and geologists lined up who could easily see and explain the flaws but they refused to listen so instead I wasted two years of my life simultaneously trying to make it work while finding increasing evidence it never would. Even after I left they still seemed sure they could make it work, but I’ve seen no signs they have. It really turned me off theoretical ecology.

  3. Stuart Smith says

    “Why haven’t they invented calculus yet?”

    You’ve had a semester of marking… Imagine the workload for a spider professor as he marks his math class! He’d never get caught up when every family is sending hundreds or thousands of kids off to college. Do spider profs even have tenure to look forward to? Spider demographics simply doesn’t lend itself to a robust system of higher education.

  4. komarov says

    I can’t get over the “complex landscape” argument. Even ignoring variety (as they did), how could anyone possibly get the idea that aquatic “landscapes” were somehow simpler than non-aquatic ones? Just five minutes of BBC footage (narration by David Attenborough optional) should be enough to dismiss the notion with prejudice.

    Heck, I might as well publish a rebuttal claiming that fish are our intellectual superiors because the live in a three-dimensional world, being able to swim around and above obstacles, whereas us surface-crawling landlubbers* are stuck with two or two-and-a-half at best. Headlines, fame and fortune, here I come!

    On second thought, meh.

    *Birds are only part-time 3D and therefore stupid, too

  5. says

    Bingo, you hit the nail on the head. Easy-peasy explanations like this for the Cambrian Explosion are absurd. There is a similar phenomenon in explaining how humans became so human. There is book after book and paper after paper, each touting one simple human need, or one simple human ability, that supposedly explains our reasoning and sociality, And what is this magical need, or magical ability? Why, it’s different in each of those books and papers!

  6. robro says

    So much emphasis on visual perception. I’m no biologist, but I smell a bias. It seems obvious that some ancient ancestor might get eaten while keeping a sharp eye out for predators if they didn’t hear or smell them coming from behind. As I understand it, sound does quite well in the water, and I believe I’ve read that chemicals are also important signals in the water.

    So, listen up folks, put your noses to the grindstone and do a little more simulating. And remember, a simulation is just an “imitation of a situation or process”. GIGO.

  7. Artor says

    I would love to host a mantis shrimp, but I can’t afford a bullet-proof aquarium.

  8. unclefrogy says

    that is what I thought as well. sight is not as important in water as smell and as we learn more sound.
    I am also kind of board with fish becoming land animals as if they were the first things to do that. I am pretty sure there were crustaceans at least all over the place. probably gastropods and insects and worms as well
    making a software simulation that changes over time is a remarkable accomplishment in it self, it may even offer insight into how things evolve but it ain’t a silver bullet and the one answer we have all been searching for, for crying out loud.
    uncle frogy

  9. stroppy says

    Psychedelic shrimp with cosmic eyeballs transcends puny calculus.

    Hypnotoad move over.

  10. consciousness razor says

    From the paper:

    For both of these computational experiments, the predator was designed as a reflex agent (aggressive prey pursuit with some randomness) with a belief distribution reflecting the likely locations of the prey when it was out of view. The prey was configured to have either habit-based action selection, or plan-based action selection with a preset number of states that it could forward simulate. Note that we were constrained to simulate planning in only the prey due to the high computational burden of simulating planning in more than one agent21; the choice of prey rather than predator was arbitrary, and we do not expect it to affect our primary findings.

    I don’t understand why that wouldn’t matter. If the idea is supposed to be an arms race of some kind, but one of the combatants is being handicapped and not actually competing in that way, because you weren’t able simulate it, then to me that sounds like you should expect it to affect your findings.

    I mean, even if you really can’t wait to publish before you’re able to make a better simulation, you should probably still expect it to affect your findings, no? Maybe I missed something…. Do they give any reason why they have those expectations? Or is it just pure optimism?

    First, we simulated midwater aquatic conditions (Fig. 1a) where the prey’s visual range was varied in a simple (open) environment. Structured aquatic environments such coral reefs are also considered in a subsequent analysis. Second, we simulated terrestrial conditions (Fig. 1b) by adding obstacles until a predetermined level of clutter density was reached (quantified by entropy (Eq. (2)), see Methods). Unlike the aquatic condition, visual range was limited only by the presence of occlusions, where the prey and the predator could not observe each other if an occlusion existed on the ray between them.

    First, it’s laughable to think these environmental simulations are even remotely adequate, and forget about “realistic.” Calling them “aquatic” and “terrestrial” and so forth is just ridiculous. They can’t even put spherical cows in their environments. If they had been able to do that, they could have called it “my kingdom for a spherical cow.” (Sorry, but a 1-sphere using a non-Euclidean metric just doesn’t count.)

    And it’s pretty wacky to toss in some arbitrary “safety” square, where the prey “needs” to go to “win” this little battle, which somehow means it’s “safe,” even though it is totally unrelated to whatever structures they put into the environment to obscure vision/motion. You climb up a tree because that’s safer, or you hide under a log, behind a rock, or whatever the case may be. In the real world, that’s what safe places, routes, etc., are like. They’re not completely meaningless destination spots that have the property that they are a red box with the word ‘safe” associated with it. If they’ve ever studied the highly realistic “Pac-Man” or “Ms. Pac-Man” simulations, they could maybe learn something from how the prey needs to navigate its complex environment in a constant search for food, while also avoiding numerous predators.

    Beyond that, the reasoning is just hard to follow for me. What they’re aiming for with all this is not increased visual range or anything having to do with perception, but a different type of parameter in the model: the depth of an abstract tree, representing possible future states that an animal imagines, uses for planning, and so forth. This is not something the animal needs to see in its environment. But the idea is that one evolved as a result of the other.

    So, if they’re trying to determine whether this helps to explain differences in behavioral planning (assuming there is an actual difference that requires explanation), then why wouldn’t they also use that parameter for prey in their “terrestrial” environments as well? What would be the harm in doing that, and wouldn’t that make it easier to make these comparisons?

    You can say that those environments physically allow for better vision, but plenty of land animals still do have poorer vision than others (no matter what their cognitive skills may be like). It’s not as if each of them have visual systems which are just as good as every other. Since it’s a central part of how their explanation is supposed to work, why would they remove that variable as soon as it’s supposed to be the most relevant? It really seems like a strange choice. And if the response turns out to be “well, we couldn’t handle doing that in our simulations either,” then I just don’t see why that’s a satisfactory answer.

  11. JoeBuddha says

    I’m sure with a bit of tweaking and some better graphics it’d make for an AWESOME first person shooter…

  12. chrislawson says

    Is there any evidence that the atmosphere or oceans became more transparent during the Cambrian? How did photosynthesis evolve if there was insufficient light for vision?

  13. astringer says

    I read “Blink of an Eye” years ago, assumed it was a nice idea, but I’m no science expert, and waited for the jury to come back in. This is the frist time I’ve seen any nod to Parker’s premise: any chance of some pointers to an update? Wiki (my usual go-to) appears rather silent aside from a brief biog.

  14. blf says

    We know camels are very advanced mathematics. Calculus? Easy-peasy, did that in the womb…

  15. wzrd1 says

    That Nature even accepted it was surprising, hell Omni magazine might’ve taken it when it still existed, Nature only embarrassed themselves.
    As that paper made greater leaps than Superman did in the comics! And of course, cephalopod vision just got a hand wave.
    And worse, it seems to have survived peer review. I miss the old days, when peers would review and savage such nonsensical works.

  16. evodevo says

    The problem is simple…I presume these people are math/computer/physics people? They deal with simple systems and have NO idea how to model actual biological systems, much less deal with multiple inputs. They assume, since they are math people, that their ideas are superior to those emanating from us lowly bio people (believe me, I have experienced this in real life) and that their setup is the most exciting innovative concept to ever be applied in the field of biology, worthy at the least of unconditional praise and acceptance…
    We’ll see where hubris gets them…

  17. blf says

    “Mantis Shrimp onna stick!
    Mantis Shrimp onna stick!
    Get ’em while your luck—OUUUGH! Arrgghhhh…“—CHOMP—”…help


  18. DanDare says

    I feel like such a failure. I have done dozens of these sandbox type simulations to muse about some aspect of evolution and never thought to publish with overarching and far reaching conclusions.
    Probably a good thing. I did a sim to test the idea that genes that produce sexual attraction end up predominantly in the kinds of body they wohuld be attracted to. I made huge assumptions about granularity of genes, abstraction of body form to rgb colour and so on. Could you imagine how that would play out in the popular press?

  19. ANB says

    “Epistemic humility just never sells.”

    Well, I’m sure you’ve noticed, but humility just never sells.

    Which: Trump.

  20. Rob Grigjanis says

    evodevo @22:

    I presume these people are math/computer/physics people?

    If only there was some way to find out, like clicking on the names in the linked article.

    Ugurcan Mugan
    Center for Robotics and Biosystems, Northwestern University, Evanston, IL, USA
    Department of Biomedical Engineering, Northwestern University, Evanston, IL, USA

    Malcolm A. MacIver
    Center for Robotics and Biosystems, Northwestern University, Evanston, IL, USA
    Department of Biomedical Engineering, Northwestern University, Evanston, IL, USA
    Department of Mechanical Engineering, Northwestern University, Evanston, IL, USA
    Department of Neurobiology, Northwestern University, Evanston, IL, USA

    Yep. Nothing to do with biology.

  21. Rob Curtis says

    At least they weren’t physicists this time! Whew.
    I’m gonna read some QFT articles now.

  22. chrislawson says

    Re: epistemic humility:

    I spend a lot of time grilling students on how we know what we think we know. Some take to it like a duck to water. Some get frustrated that we’re not learning “core” clinical medicine. Over time, though, even the ones who just want to learn what they think they need to know eventually come to an understanding that the medicine I was taught has been about 50% superseded, and the same will happen to them. It is critical to have the appraisal skills to work out what ideas need to abandoned and what ideas need to be maintained when the evidence changes. You can farm this out to other people to some extent by deferring to consensus guidelines, but this doesn’t always work — particularly when different expert groups have divergent guidelines — and even when it does work, it means you will always be 2-5 years behind the research.

  23. chrislawson says

    Re: epistemic humility part 2:

    Evolutionary theory is one of the most powerful models ever developed and, despite the inevitable controversies and knowledge gaps, is the only extant theory capable of explaining the diversity of life on the planet. However, it doesn’t follow that any given evolutionary hypothesis is automatically as robust as the fundamental theory. It’s really the same shell game the banks were playing prior to the subprime mortgage crisis.