Guest Blogger: Mike Vorhis, author of ARCHANGEL suspense thriller and OPEN DISTANCE adventure thriller

Figure 1: Two Midges
Figure 1: Two Midges

This post is more of a theory—a postulate—than it is a claim of pure fact. I’ll get right to the point: I’ve come to believe that the most important element of a trout fly is not its overall realism but its presentation of a “trigger feature” or two. This is in line with a couple of other articles posted here over the past year, and is also at odds with some widely held misconceptions. Let me explain why I think this:

Your eyes catch a glimpse of something in a chair at a Parisian outdoor café. It’s leaning against the chair’s back rest, a steaming cup of cafe on the table before it. A hat (a beret, naturally) rests on top, fashionably cocked to one side. What is it you see?

Or…baseball glove on, you’re standing ready in deep center field, facing the infield; squinting into the low evening sun. You hear a ‘crack.’ A moment later, accompanying faint yells emanating from the opposing team’s bench and a cry of “Two two two!” from your catcher, a fast-moving shadow seems to be coming directly toward your head. Do you raise your glove to make the catch?

Nope, it was a tall backpack on the café chair…but then I only gave you a couple of isolated bits of data, and you did the best you could in matching them to tidbits stored in your head. And nope, it’s a redwing blackbird swooping toward you, flushed from the grass by a wildly hopping grounder making a bid to get past you to your right…but again I only gave you a couple of pieces of info and you did the best you could.

Identifying something can be a process of matching small isolated details and then drawing a hasty conclusion.

In modern “deep learning” computational machines, large quantities of discrete data items are stored in advance, to feed a decision-making algorithm. The other main inputs to that algorithm are instantaneous observation of a complex scene, and a priority. Take for example a self-driving car system; it analyzes a moving scene fed to it via a live video stream. In any given handful of video frames, it compares what it “sees” against millions of isolated data points pre-stored in fast, tightly coupled memory. It compares rhythmic motion of linear ‘things’ at baseline (ground) level, and calculates a rough proportion and speed, to determine whether those are legs and that thing is probably a walking human being. It then applies a very high priority to the task of not running into that thing—higher priority than if it had decided that the thing was a tumbleweed.

We think the same way. For example, we can barely make out visual information through dense fog, yet still, by a few basic pieces of the scene that we match against info in our memories, we can decide there’s a person standing or walking on the roadside. It might in fact prove to be a mannequin or animated human-shaped sales display, but we draw early conclusions with the limited data we have…and then act on them.

Our pre-stored data can be innate or learned. Without ever having to be taught, we know instinctively what a large predator’s silhouette could look like, and our reactions sometimes draw such an early conclusion if we’re surprised…but yet if one piece of data we have stored is that last week we saw an operational oil rig along this stretch of road, then in a fog we can swiftly conclude that that strange lurching shape up ahead is not a dinosaur.

Every creature with a central processing system (a brain) does this. We know trout are unusually adept at pattern-matching; as expected, their brains seem to specialize in that task, rather than toward the many other ways a brain could work (such as socially or to build structures or use tools). But due to the size of the computational task their brain must perform and the small size of the brain itself, the pattern-matching task must be greatly simplified. Enter those simplistic, discrete “trigger features.”

Many—I’m going to say most—of our best trout flies are not faithful replicas of insects or aquatic critters. How many dry flies use wound hackle comprised of hundreds of barbs to represent the handful of legs a flying insect actually has? How many flies bother to hide the hook point at all? Instead they use a feature or two that triggers “matches” in a simple creature’s brain. A bit of red. A long tail (and too long works fine, since trout seem to care more that it’s just plain long than that it should be 2.78 times the length of the abdomen). An eye. A dark thorax. Fuzzy. Light olive color.

Figure 2: Two Hoppers
Figure 2: Two Hoppers

Put just a couple of these features together on a fly and, if they get a ‘hit’ in a trout’s memory banks, you’re likely to get a hit on the stream. You’ll get more takes than you will on a detailed stamped-out plastic department store bug that looks “exactly like the real thing” to us.

That the pattern-matching task must be simplified for the weakness of the trout’s computational engine means their conclusions are prone to considerable error. They do well with what nature throws at them, but we’re cheating by creating non-food which still passes their simplistic inspection. No one gave us the code but over several centuries we’ve “reverse engineered” their algorithm through trial and error, to the point that we now know a bunch of “trigger features” that seem to work, along with the conditions under which they do. We have a lot yet to learn, of course.

Particularly absent in their ability, in my experience, is counting precision, except perhaps for noting rough proportion. Remember the film “Jeremiah Johnson,” based on the Vardas Fisher novels “The Crow Killer” and “The Mountain Man”? Hiding behind their horses as they stalked a herd of elk, Jeremiah asked the character played by Will Geer whether the elk would see their legs and spook. The old man scoffed, “Elk don’t know how many legs a horse has!” Well, trout don’t count insect appendages and in my opinion don’t normally measure wing and tail lengths very precisely either; general proportion and general expected size is usually good enough.

The decision-making algorithm of a trout seems to involve many possible pieces of data, yet still be quite simple. It seems to say something like, “that’s too big,” or “there’s no dark brown thing like that in my stream,” or “I see yellow on the bottom and black antennae…that’s all I need to know.” It would be as if you were to see something on your kitchen table and say, “It’s orange…and there’s something thin sticking out the top…so it’s a pumpkin.” You matched two trigger features against your knowledge base and drew a conclusion. (That it’s actually an old tennis ball into which your kid stuck a nail would become clear when you try to make a holiday pie out of it.)

The opinion I’m expressing does NOT mean trout don’t think! We simply have to define what thinking is, and be a little honest. I believe trout, as do almost all animals, weigh data points and observations in a manner similar to ourselves, although with considerably less processing power and thus far less complexity. Every thinking engine, be it an electronic neural network processor or a biological brain, attempts to fit stored elements of data against in-situ observation to draw a conclusion. If the data points are few and the conclusion-drawing algorithm simple, well, you have yourself a rather simple thinking machine, and one probably weighted more toward yes/no pattern matching than toward theoretical extrapolation. But if even a single simple conclusion is drawn, regardless whether it’s “this is a step in the solution to Fermat’s Theorem” or “this is not food,” then make no mistake, thinking has occurred. It’s just a question of degree.

Of course humanity has always been quick to reserve the words “reason” and “thinking” for ourselves; even other mammals—deer that deliberately begin to avoid a farmhouse out of distrust for a new inhabitant, or higher primates who demonstrate advanced puzzle-solving ability are one and all denied their due by many of us.

I choose to not be unduly dismissive of trout. Whether their stored data is born-in (we call it instinct) or stored through a process involving personal experience…such as growing shy of chartreuse fly lines or the same pattern floating by, season after season…matters little. It’s data, and it feeds a simple decision-making algorithm.

Figure 3: Classic Fly Collage
Figure 3: Classic Fly Collage

So when I open my fly box, mid-current, to pick a pattern, I think about what “features” I might want to present. I shy away from perfectly detailed things involving exactly six-point-zero legs like I avoid molded rubber hellgrammites. Instead I choose whether I want to show fish a bug’s long tail, or a red hot spot, or a wing case with a splash of pale yellow that implies it might be splitting open. I think in terms of isolated features—try to match a couple of discrete data points that make sense for the season and time of day…then present it faithfully per the current. I believe discrete fly features trigger trout into a decision in my favor more than would an overall exact replica.

Am I right? Well I have no absolute proof other than the plethora of “attractor” and vaguely impressionistic patterns that look like nothing to us but still draw savage strikes, including the inspired work of the great tiers and so many deadly old-time wet fly patterns that are shaped like nothing we know but still slay by virtue of a color pattern alone or a simple fin-shaped bit of feather. I do know that thinking about it the way I describe here certainly improves my fishing success.

9 Comments

  1. I think Mike is spot-on with his emphasis on triggers rather than ultra-realism. The late Gary LaFontaine also put forward this theory, and had he not been taken from us in such an untimely manner I think he would have developed it further. I think that a lot of fly fishers are way too anthropomorphic with fish, trout in particular. When we fail to catch them, we think it’s because they’re “smart.” We attribute to them all sorts of anayltical thinking that is totally beyond the abilities nature gave them. Trout do not think. They respond to stimuli–triggers if you will. Analysis and critical thinking take place in the brain’s frontal lobes, and this part of a fish’s brain is so small as to be neary non-existent. Fish do not run down a checklist of whether a fly accurately represents all the features of the natural. If your imitation hits just a few of the right triggers (which includes the behavior of the fly as well as its appearance), they will readily ignore the absence of features that a human thinks should be there. They will also readily ignore features that should not be there, such as a protruding hook bend and point or a huge number of “legs” in the form of a bushy hackle. Ultra-realistic tying is a fascinating part of the fly tyer’s art, but it is assuredly not essential to the practical craft of tying effective fishing flies.

  2. I agree with Mike and in my experience movement, be it the very subtle movement of peacock hurl or larger movement of hackle fibers, is very important. I do however want to add something to think about. After teaching fly fishing for a number of years I stumbled upon a video clip of adult trout taste testing all kinds of stuff in the drift spitting out most of it and eating what they deemed edible. Everything I thought I knew about picky trout disappeared. They were mouthing twigs and leaves among other things, stuff with no moving fibers at all. Since viewing that clip I have come to believe that there is a piece of the “What do Trout Eat Puzzle” that we don’t understand. It would be great to hear more discussion on that.

  3. Thank you Mary, for those thoughts. I didn’t realize that Gary LaFontaine had espoused the trigger feature theory per se, but I did know that his ties were excellent examples of it. many, may other great patterns go there as well.

    But make no mistake–trout do “think.” They run an algorithm. It’s just a relatively simple one. What they see is an input to that algorithm; the rest of it is a search through their data banks. Whether that search is random or sequential we cannot know and it doesn’t matter–it only matters that it happens. And finally there is a conclusion/decision part of the algorithm to finish it off.

    All creatures with processing engines (brains) and sensory capability of any kind do this. That’s what brains are there for. If the frontal lobes are small, the processing engine is small. That’s all.

    As you note, trout are capable of ignoring the presence of fly aspects that should not be there, and equally capable of ignoring absence of fly features that would otherwise get a “data hit” in their brains. Why they sometimes take a given fly and sometimes do not is also unknown by us, except to say that, like humans, decision-making processes are not an exact science–they are applied individually. Maybe one specimen will ignore the absence of a wing case but the next one will be a little bothered by it. We don’t know in detail. This is probably why some patterns kinda work and others are absolutely killer.

    One has to define “thinking.” If you limit it to human-like decision-tree analysis that results in the penning of novels and the design of computers, then it’s true that trout have not been known to do these things…there’s no call for them to. If you define it more rightly as the processing of input data, the comparison of it to stored data (using even a random search-and-compare), and then the drawing of a conclusion, and leave it at that, then they absolutely do. They must.

    And we rely on them doing so; we wouldn’t have it any other way. otherwise we’d not bother to study entomology; we’d just throw the same pattern at them forever and some of the stupid ones would eventually grab it.

    When I was a child we raised guinea hens (wish I had those feathers now!). They would not step over a garden hose. They may not have known why (just as people don’t know why they’re afraid of spiders or dark shadows moving across the floor). But something in their brains said “this matches a fact in my head and it’s dangerous and I’ll be damned if I’m going to step over.”

    A radio talk show host once said he raised goats at his home, and that they would freeze motionless if his wife, going out to feed them, happened to be wearing her new pair of leopard-pattern slippers. The goats had no idea why…but the pattern matched something in their stored data that said, “this is dangerous,” and so every goat, on seeing those slippers, froze like they were dead…to save themselves.

    And lest we label that as automatic stupidity of “non-thinking” creatures, remember that human infants (who, since they’re our species, we’re willing to admit have the capacity to think) grow instant and undeniable attachments to anything with a face, even if the eyes are sewn-on buttons. They see a face shape, they run a simple algorithm, the “face” matches something in their stored data, and they decide to befriend it.

    It’s so interesting to me that we still insist on defining “thinking” as the very flavor of intelligence we happen to possess. I doubt we’d fare well in the dolphin/whale world’s assessment of it, given that it would be based on sonar processing and telepathic capability. I think we come up well short of the ability of swifts and terns to calculate a position in three dimensions in real time and to physically maintain it with millimeter precision. I suspect dogs and wolves and foxes must roll their eyes at our inability to predict earthquakes and to sense emotions of those to whom we’re bonded…even across species. I imagine horses try their best to overlook our lack of a noble, ethical, dignified gene.

    In my article I rightly defined “thinking” in the most inclusive, most simplistic way–the processing of a bit of data by comparing it to another bit of data stored in memory, and the drawing of a simple conclusion. I defined it simplistically not for benefit of other species, but quite frankly so as to remain inclusive of us. Please, please let us be the generation who finally and definitively abandons the notion that we’re the only ones in the cosmos with some processing power–some “thinking” ability. Yes, we have some. Our type is our trademark. There are so many degrees of it, and so very many other types.

    All that said, as Mr. LaFontaine and many others have also seen, expecting trout to make their decisions simplistically is quite sufficient for a fly angler’s purposes. So without being dismissive we can most certainly just focus ourselves on simple, broad-stroke things. Furthering our theories about their processing algorithm is a fascination we all enjoy, and over the centuries it has paid off incrementally. But most of the details of how they process input data and draw a conclusion on which to act will always remain a fascinating but elusive hobby we pursue.

    See, now, that word “thinking” must have been a “trigger” for me. : ) Thanks again Mary.

    Phil, I’m with you on the supreme importance of movement (of hackle, guard hairs, the entire fly…whatever). And your taste-testing input is fascinating and I’m not a bit surprised. Could they be calibrating tastes? Sensing the season by the plant material coming down? Getting vitamins, like a carnivore does when it eats grass? Something else? You’re right, there’s so much we don’t know. Tasting would definitely add some kind of data to their equation. Fascinating.

    – Mike

  4. Point taken, Mike. It all depends on how one chooses to define “thinking.” Do trout and other gamefish take in and process information. Of course. I suppose I should have said that they do not think “like humans do.” LaFontaine spoke of both positive and negative triggers influencing a trout’s decision to eat or not to eat. Even the most primitive organisms can learn to avoid unpleasant stimuli. There’s a classic behavioral science experiment in which aquatic flatworms are placed in an aquarium in which they receive a mild electrical shock when they enter certain areas of their enclosure. They very quickly learn to avoid those locations. Does this very primitive animal think? Again, it all depends on your definition. It’s all relative and a matter of degree. We fly fishers take in a lot of empirical data in the course of our fishing experiences. I think that we often draw unjustified conclusions based more on intangible feelings and hunches than on logic. There’s nothing wrong with that. It’s part of the enjoyment of the sport, and in fact it’s inevitable. There are far too many uncontrollable variables involved. Fly fishing is not nor will it ever be a science, and thank heaven for that!

  5. We’re of a mind, Mary. It comes down to semantics, degree and a little humility (since as Phil pointed out, there’s still far more we don’t know about other critters’ behaviors and possible cranial processes than what we do).

    > Even the most primitive organisms can learn to avoid unpleasant stimuli.

    Yes, and often it’s data not even learned y them–it can be learned thousands or millions of years before and simply born in. Still data though. As an example, there are many preferences and many “facts” I carry around with me and yet have no idea why I believe them or where I got them. But it’s still data.

    Nor is your earlier point lost–assigning complex cleverness credit to a trout we can’t seem to fool may be seriously unwarranted. Suffice it to say they’re a black box to us and we’ll never know what, exactly, goes on in there. I think when we declare a cagey old fish to be a genius, it’s mostly to absolve and compliment ourselves.

    And, dang it, I’ll never give that up. 🙂

    – Mike

    PS. I love trading viewpoints on these hard-to-nail-down, many-angled topics!

    1. Yes, I have enjoyed this exchange too. By all means, never give up giving trout more credit than, in an objective sense, they might actually deserve. There is a definite need for some soothing balm for the angler’s bruised ego at times. We really don’t know what goes on in fish’s brains, and I doubt we ever will. Our opinions can be informed somewhat by science, but it will never result in conclusive answers. It’s fascinating to speculate and form theories. However, it really doesn’t do to take ourselves too seriously in these matters.

      1. Well I for one am not going overboard. I’m going to tie up a few of those relativistic patterns Einstein used to tie…what were they called…the ‘Quantum’ something-or-other…see if I can finally fool the big extra-smart ones. That fly is mostly some kind of grey matter, with Latin scribings on the thorax. Yes, I’m sure that’s what is needed, for those smart ones….

Leave a Reply

Your email address will not be published. Required fields are marked *