Introduction

Artificial Trout Flies: Spoilt for Choice or a Marketing Miracle?

"Although suggestions abound for what artificial lures an angler should use under what conditions, few or none of these recommendations are based on experimental records. Most seem to be somewhere between witchcraft and snake oil. But, then, those promoting the use of one fly or another are often attempting to catch anglers, not trout." (Grubb 2003 p171).

"If our focus remains stuck on creating flies to solve business problems instead of fishing problems, it’s a safe bet that those solutions will be a long time coming."  (Juracek 2020).

"In my opinion ninety-nine per cent of the new patterns we see are completely superfluous and they do fly fishing a disservice by confusing beginners and obscuring real innovation." (Herd 2003 p358)

This is not a modern problem; writing in 1870 Henry Cholmondeley Pennell dismissed as useless what had been written in the preceding 200 years about what flies to use, and how to construct them. Consequently: "it is nevertheless true that nine-tenths — or rather ninety-nine hundredths — of these graceful combinations of furs, silks, and feathers represent so much wasted time, money, and ingenuity" (p 78).

There is one simple solution if you are faced with this problem. The fly-fishing literature is sprinkled with short lists of recommended flies. You will find identical flies across many of these selections. I fish on relatively small freestone rivers, and I can say from experience that I would be happy using the flies recommended by Tom Rosenbauer (2011 p 126-133) in The Orvis Guide to Small Stream Fly Fishing.

In my opinion, Rosenbauer's recommendations are prey images that contain features (sign stimuli) corresponding to the trout's search image. For example, the Hare's Ear (in both dry and wet versions), the Pheasant Tail, the Sparkle Dun, and the Elk Hair Caddis.

Many people are content to leave it there, with a recommendation from a reliable source. Others may want a reason for relying on a short list of flies. The straightforward answer is that these flies have stood the test of time. For example, the Hare's Ear has been around since the 18th Century, and a relative newcomer Sawyer's Pheasant Tail Nymph since 1958 (Whitelaw 2015).

But if you tie your own flies, or if you are simply curious, you may want to know why these flies are more effective than others. This might help you distinguish the "completely superfluous" from "real innovation" (Herd 2003), and solve "fishing problems"  (Juracek 2020).

A good starting point is this perceptive comment from Steeves and Koch: "It might be said that all the flies ever tied to imitate hatches have been formulated in the search for the correct sign stimulus for different situations" (Steeves and Koch, 1994 p 70). This was a groundbreaking book that devoted a chapter to introducing ethological concepts. Nevertheless, until recently, the design of trout flies has been hampered by being located within a theoretical vacuum.

There is one ethological concept - the supernormal stimulus - that has received relatively little attention in the fly-fishing literature. Kenneth Boström's fly, the Rackelhanen, was an important step in the evolution of fly design that inspired several fly-tyers to break free from traditional English hackled dry flies. It led to Bob Wyatt's 2013 book  What Trout Want  that based fly design on the discoveries first made by ethologists in the 1950s.

---

The Rackelhanen - inspiration for super-stimulus trout flies ?

Rackelhanen available from  Luke Bannister.

The Rackelhanen was developed in 1967 by Kenneth Boström. It is constructed from just one material - polypropylene yarn.

It's worth reading his  detailed account  of how to fish this 'ugly duckling' -

"free floating as a dry fly"

"Stripping on the surface. Short and "nervous" pulls, 2 inch long, with a short pause between them. "

" Pull the line so that the fly drags under the water surface... Make a short pause, and the fly floats up again"

A Black Rackelhanen is an excellent colour for Dartmoor.  Luke Bannister

Boström described the Rackelhanen as " .. not very beautiful, it's appearance was almost frightful." And this view was initially shared by John Goddard (2002 p51-54) when he was introduced to the Rackelhanen on the Kennet (an English chalkstream) : "I must admit I was very unimpressed when he  [Preben Torp Jacobsen]  first showed me this new artificial, as it looked just like a lump of dark brown wool thrown on a hook... So far as at a casual glance I could tell, it really bore little resemblance to any form of natural food."

Goddard's attitude that trout flies should imitate natural flies is evident in this BBC TV programme he made in the 1980s with Brian Clarke.

Goddard quickly changed his mind when he "..caught a lot of trout on this nondescript pattern.."and it had "accounted for several large and very difficult trout that would not even look at any other pattern so I am now absolutely convinced this is a winner."  Goddard went on to model his Poly-Pupa and Poly-Caddis flies on Boström's Rackelhanen.

Luke Bannister recommends and supplies Grey, Olive and Black Rackelhanens for West Country rivers; his Yellow Sally fly is tied Rackelhanen-style details here

Before he met Kenneth Boström, Hans van Klinken the inventor of the Klinkhåmer Special, "tied flies in the traditional English shoulder hackle flies".  He had this to say about the influence of the Rackelhanen on his fly tying: "It was the “Rackelhanen” that set me free from old traditions , made me innovative and allowed me to think differently. In Scandinavia the “Rackelhanen” is still a very popular fly  [e.g. Krogvold]  but worldwide this fly has never got the attention it surely deserves. I have no idea how my fly-fishing would look today without the discovery of the Rackelhanen but that wonderful sedge imitation gave me enormous self-confidence and inspiration to start a completely new way of fly-tying. " (van Klinken 2017) [emphasis added]

Perhaps unknowingly, Kenneth Boström had handed Hans van Klinken the key to a door that had been shut by  Frederick Halford , and then locked by his followers, in the late 19th and early 20th century.

The Klinkhåmer Special with its prominent wing and sunk abdomen, fulfills the  two cardinal features  of an effective dry fly identified by Marinaro, as well as Clarke and Goddard:

penetration of the trout's mirror at a distance,

and gradual appearance of wings as the fly approaches the edge of the trout's window.

Hans van Klinken's Klinkhåmer Special was the inspiration behind Bob Wyatt's Deer Hair Emerger [DHE]:  "Unlike a high floating dry pattern, the trout notices a fly like a semi-sunk Klinkhamer Special or my own pet emerger, the DHE, [Deer Hair Emerger] from a greater distance and locks onto it. These flies share the strong triggers of a nymph and the hi visibility of a dry fly. The trout completes its behavioural response to a potential prey item by eating it. To a trout, charged up and in feeding mode, the fact that the fly doesn’t look exactly like the other bugs on the water is outweighed by its built-in behavioural releasers. " (Wyatt 2012). It is considered to be a multi purpose emerger pattern that works equally well on caddis and mayfly hatches (Rakkenes).

Wyatt's flies are probably the first commercially available modern trout flies designed on a scientific basis:  "Behavioural science terms like ‘behavioural releaser’, ‘supernormal stimulus’, ‘optimal foraging strategy’ and ‘fixed action pattern’ entered my angling vocabulary. Everything just sort of came together and for the first time in my angling life started to make sense." (Wyatt 2012).

The success of Wyatt's flies may be due to the simplicity of their construction, and use of widely available tying materials, which makes them easy to tie ( Hans Weilenmann undated), and overcomes the problems encountered by earlier innovations such as  Marinaro's Thorax Dun , and Austin's  Tup's Indispensable.

The rest of this essay explores the possibility afforded by these ethological concepts to the design of artificial trout flies.

The next section is an appreciation of much earlier flies. Frank Sawyer's Pheasant Tail Nymph shows how observation and insight into the structure and behaviour of insects and trout, created an outstanding trout fly that exemplify an 'ethological' approach to fly design.

This is followed by examination of Skues Tup's Indispensable and Carrot fly, flies that anticipate the lessons provided by Sawyer's Pheasant Tail Nymph.

Then the essay explores the possibility that the attractiveness of these artificial flies involves sign stimuli, and can be explained in terms of the ethological concepts - supernormal stimulus and heterogeneous summation.

---

The ubiquitous Pheasant Tail Nymph

We can apply an ethological approach to explore the possibility that several flies made from pheasant tail fibres incorporate sign stimuli that elicit feeding behaviour in trout.

Frank Sawyer wasn't a trained ethologist, but an ethologist would have appreciated his approach to designing nymphs on the basis of observation, and the simplicity of features selected for inclusion in the 'prey image' . "The main features which were obvious to me when watching mature nymphs, were the swelled wing cases and the use of the tail for propulsion, coupled with the general stream-lined effect. (Sawyer & Sawyer 2006)

Sawyer introduced the Pheasant Tail nymph - a simple fly construced from copper wire and dark pheasant tail fibres to imitate Baetis nymphs - which is cast upstream of the trout so it sinks to a trout's level.

In this 1950s video, Frank Sawyer gives a demonstration on how to tie his nymph and a dry fly. This is the only known surviving footage of Frank Sawyer actually tying a nymph.

He is also remembered for introducing a way of fishing this fly. The rod tip is lifted so that the fly ascends in the water in front of the fish- the 'induced-take' technique. Sawyer's name is often associated just with the induced-take, but he also stressed that a fish could take a nymph as it sank, or moved from side-to-side: "a much greater attraction is when the nymph is made to check in its descent and start to move to one side or the other, or upwards, as though swimming...The rod movement to impart animation should neither be fast nor jerky, but merely a gathering of all slack with the line control hand, followed by a slow even lift of the rod tip with just sufficient speed to make the nymph swim. "

Actual flies tyed by Frank Sawyer

Sawyer (1979) commented: "General shape and colouration, together with the right size is of greater importance than an exact copy. My two universal patterns, as I call them, are the Pheasant Tail and the Grey Goose. The Pheasant Tail serves for the darker coloured nymphs and the Gray Goose for the lighter ones." Sawyer (1979)

Frank Sawyer stressed the importance of designing his nymphs so that they could be fished at the  correct location  in the water column: "Though there is importance in having the right colouration and size it is the general effect seen by a fish that counts...one of the essentials is to construct a nymph pattern so that it can penetrate the surface and sink to the level of a feeding fish. "

The effectivenees of simple flies constructed from pheasant tail fibres is not restricted to English chalkstreams.

Arthur Cove's Pheasant Tail was developed to imitate 'buzzers' (chironomid midges) on Eyebrook reservoir.

The American Al Troth based his Pheasant Tail nymph on Sawyer's original pattern but used peacock herl as thorax material.

The Teeny Nymph is another example of a simple but effective trout fly which may imitate a shrimp. Size and colour variations of Jim Teeny's basic pattern have been responsible for catching 25 IGFA (International Game Fish Association) fresh and saltwater world records.

The simplicity of these flies suggests several candidates for sign stimuli that elicit a trout's feeding response:

movement - these patterns tend to be fished with some form of movement

colour - colour is often varied to match the colour of the natural nymph

thorax - is present but construction materials vary (Sawyer and Teeny used pheasant tail; Cove used rabbit fur and Troth used peacock herl)

body shape - designed to represent shape of natural (Sawyer and Troth tied a straight body to represent a mayfly nymph; Cove tied around the hook bend to represent chironomid pupae)

the ratio - between body size and thorax may be important when representing particular insect groups

This table presents the design elements in several 'classic' artificial flies used for sub-surface presentation to trout in rivers and stillwaters.

All of these successful classic trout flies have the following design elements in common:

body made of pheasant tail fibres

thorax made of pheasant tail fibres

movement imparted by the angler

Application of Lloyd Morgan's Canon would suggest that an artificial fly constructed with a straight body from pheasant tail fibres and some form of thorax which is moved in the water should catch trout. It is interesting that the flies constructed by Sawyer, Cove, Troth and Teeny are more elaborate than this simple pattern. For example, Sawyer's nymph has a tail.

However this analysis does not discount the possibility that the tail is a sign stimulus when trout are feeding selectively on mayfly nymphs. Likewise, the curved body in Cove's fly may be a sign stimulus when trout are feeding selectively on chironomid pupae.

Selective feeding may be the result of the operation of a  search image  consisting of several sign stimuli.

Thus several different sign stimuli may act together to trigger the trout's feeding behaviour. Trout may 'add-up' sign stimuli to determine if an object is edible.

This would be an example of the  law of heterogeneous summation  which predicts that incorporating several sign stimuli into an artificial fly could increase its attractiveness to trout. In fact, the law suggests that these artificial flies could be more attractive to trout than the natural insects they are supposed to represent.

---

Oliver Kite's Bare Hook Nymph

The writings of several experienced anglers suggest that the 'sign stimuli' involved may be even simpler:

Shape: the outline of a nymph represented by a thorax composed of a few wire wraps

Movement  of the 'model' in the water - the 'induced-take' technique

Sawyer Pheasant Tail nymph

Oliver Kite, in his 1963 book Nymph Fishing in Practice, commented that he continued to catch fish on a worn, ragged Sawyer Pheasant Tail nymph  "provided that the basic structural outline with its humped thorax remains."  (Lawton 2005 p 87)

Kite's Bare Hook Nymph from Sutcliffe (2016)

Oliver Kite took this observation one step further. He enjoyed success with this ' Bare Hook nymph' which consisted of a few turns of wire wrapped around the hook shank. He was also able to catch trout on this simple nymph whilst blindfolded by using Sawyer's 'induced-take' technique.

Roy Christie (nd & personal communication 2021) takes up the story: "A disciple of Sawyer, Oliver Kite fished the same water on the Avon and developed the Sawyer nymph to minimalist proportions. Kite reduced the pattern to the point of removing the herl, then the wire and caught trout and grayling by induced action on bare hooks. When he reported this to Sawyer, he was allegedly informed that without any dressing the hook could not be a fly and that he was thus outside the rule book of the Club. Thereafter Kite would attach a knot of silk or a turn of wire to remain within the remit. Hence the infamous bare hook nymph."

Roy Christie's Copper Wire Hare's Mask

Inspired by Kite's success, Roy Christie (nd) developed his Copper Wire Hare's Mask fly with which he has "..spent many hundreds of hours using this system and caught thousands of trout with it." But he adds:" Does it always work? Well, no."

Lawton (2005 p 89) recounts an interesting anecdote about Kite's habit when he met an angler for the first time; he would  "put a hook in his fly-tying vice and pluck a lump of wool from the carpet, or from the sweater he was wearing, and then tie it to the hook as a simple thorax. With an explanation of "that'll do" it was off to the river where fish would be caught on the 'carpet' or 'sweater' nymph."

Raymond Baring also found that a Pheasant Tail nymph increased in attractiveness as it became more and more bedraggled, and finally lost all of its original dressing

Ed Zern (1979) described how he caught trout on a pheasant tail nymph that was " a bare size 18 hook with three turns of fine copper wire around its short shank and nothing else - no fur, no feather, no silk, no tinsel."

Lawton (2005 p110) recounts a meeting between Lee Wulff and Sawyer in 1972 when they discussed the similarity between each other's nymphs. Sawyer remarked "I have to dress them up like this to sell them". Wulff had the same problem with his version of the Nondescript nymph to which he added a few turns of peacock herl at the head to "dress it up a bit". Wulff observed that "It is apparent that most anglers buy flies for their own enjoyment rather than that of the fish".

Procter's Whisper Nymph from Procter (nd)

Paul Procter (nd) developed a series of Whisper Nymphs based on Oliver Kite's bare hook nymph. He commented: " believing in scantily tied nymphs takes some doing, and that's the very secret, faith! ... never question the trout's ability of detecting minute flies, fish have excellent vision, "

---

North Country Spiders video by Robert Smith

North Country spiders   are similar to the minimalism of bare-hook nymphs. The use of peacock herl or fur to form a thorax is consistent with this idea. Spiders tend to be simple flies with slim bodies consisting of touching turns of silk thread, and a soft hackle. It has been suggested that the mobile hackle represents an emerging nymph (Harding 2009).

Frank Sawyer's observations, and Oliver Kite's practical demonstrations, suggest that shape and movement are sign stimuli. In his time, Sawyer's contribution was unique. Even today our scientific understanding of the role of these visual stimuli in fish predatory behaviour is a relatively undeveloped research area, compared to studies of the morphology and electrophysiology of the fish visual system. For example, in a laboratory, fish are able to learn to distinguish between a huge variety of visual shapes on the basis of their form and orientation. But these 12 early studies (1929-1974) tell us very little about the visual discriminations that fish are able to make in their natural environment because these stimuli are not likely to correspond to visual stimuli in that environment (Douglas & Hawryshyn 1990 p393-4).

More recent research has focussed on brain areas responsible for processing visual information. Salva et al. (2014) review useful recent research into, for example, motion of prey, and perceptual constancy - "Ensuring constant perception of invariant object properties such as size, shape and color, despite the constant modification of the physical (proximal) input reaching the retina, due to changes in viewing distance, perspective and illumination conditions." An important outcome from this research is that complex abilities require much less "brain power" than previously believed.

---

Skues and the mysterious Tup's Indispensable

For me, the thing that shouts out from the design of Sawyer's Pheasant Tail Nymph, and similar flies, is the importance of shape - the hump that mimics the thorax on the natural nymphs he was imitating. But it's worth remembering that Sawyer was fishing on chalk streams, and he designed his nymph without a hackle to sink quickly so that he could impart movement to induce a take from trout he could see. We don't have Sawyer's advantage; we can't rely on seeing trout take nymphs on our local freestone rivers. Who better to ask for advice than Mr. Skues ?

George Edward MacKenzie Skues is famous for developing nymph fishing on chalk streams. Described as a "mellow nonconformist" (Hidy 1979), he gave the self-effacing title "Minor Tactics of the Chalk Stream" to his ground breaking book published in 1910 that challenged the prevailing view that the angler should only cast to rising trout, and only use a dry fly floating on the surface of the water to represent a dun.

Skues pointed out that trout do not feed exclusively on duns floating on the surface. Nowhere is this more true than on freestone rivers like the Devonshire Avon where days can pass without seeing a floating dun. Our trout probably do most of their feeding on the bottom, or mid-water where they capture ascending nymphs. Our 'rising' fish may in fact be taking nymphs that are in the process of breaking through the barrier between air and water and shedding their exoskeleton to emerge as a dun.

Therefore Skues' tactics are ideal for the conditions we face.
Skues described four flies he tied for use on the Devonshire Avon - my local river. In 1945 Skues wrote to a Mr. S. Roberts of Exeter describing four flies he had recommended for use on the Devonshire Avon.

"The flies I tied for my friend Col. Jesse to use on the little Devonshire Avon were all rather large, being on No.2 [= today's size 13] Carlisle round bend down-eyed pattern [hook] and comprised the following:

1. Lightish dun cock's hackle and whisks. Body hare's poll.
2. Rusty blue cock's hackle and whisks. Body crimson seal's fur. Rib gold wire.
3. Rusty blue cock's hackle and whisks. Body 3 strands of pheasant tail. Rib gold wire.
4. Tup's Indispensable.

For that water he never seemed to want any others. I tied them as a guess as likely to suit that tumbling water and he did well with them all, particularly No. I. " ( Extract from "Angling Letters of GEM Skues", Editd by CF Walker, published by Adam and Charles Black, London, 1975).

Skues fly #1 closely resembes a Gold-Ribbed Hare's Ear, a  favourite fly  on chalk streams that I have suggested may be a  super-stimulus.

Skues fly#2 is interesting because the whole body consists of crimson seal's fur, and red seal's fur is thought to be an important component in the dubbing used in a Tup's Indispensable.

Skues fly#3 resembles a Pheasant Tail Nymph

Skues fly#4 is the famous Tup's Indispensable.

Břetislav Kašpar (2017) has written a useful account of the history of this fly.

It is worth describing how Skues discovered the effectiveness of a Tups when fished sub-surface, how he modified the original pattern, what he thought it represented, and what it led to.

The Tups was invented in Devon (UK) as a dry fly by a commercial fly-tyer, R.S. Austin. It's always struck me as a descendant of an earlier West Country fly - the Half-stone - which had a body in two parts, the rear two thirds yellow floss, and the front third mole fur. Skues quotes the Half-stone as a model for the shape of a nymph (Robson 1998 p85)

Skues tied the Tups with" three or four turns of coarser untwisted primrose sewing silk at the tail, body rather fat, of a mixed dubbing of a creamy pink (invented by Mr.R.S.Austin the well-known angler and fly-dresser of Tiverton) .." (Skues 1914 p 17) [emphasis added].

Tup's Indispensable nymph tied to Skues pattern by Jim Nice

I was a bit perplexed when I compared the drab body colour in this picture of a Tup's Indispensable nymph in Haytor & Skues (2005), with Skues' (1914) description of pink dubbing. Many years later, in the Summer 1928 Flyfishers' Club Journal Skues described the body as: "A variant of the Tup's Indispensable, using a greenish-brown seal's fur instead of the red seal's fur of the original pattern." (Overfield 1977 p108). Maybe in 1928, Skues was belatedly burying a crucial element - red mohair / seal's fur blended with other materials to produce a pinkish shade - to protect Austin's commercial interests.

Initially Skues used it as a dry fly, but discovered by accident that it caught trout when wet, "and, indeed, when soaked its colours merge and blend so beautifully that it is hardly singular; and it was a remarkable imitation of a nymph I got from a trout's mouth." (Skues 2014 p31).

"It was the foundation of a small range of nymph patterns, but for underwater feeders, whether bulging or otherwise, I seldom need any thing but Tup's Indispensable, dressed with a very short, soft henny hackle in place of the bright honey or rusty dun used for the floating pattern." (Skues 2014 p31)

It's interesting what Skues said, and left unsaid, about the Tup's Indispensable. He tied the fly with a "soft henny hackle". This would have allowed the fly to sink just beneath the surface to target trout seen 'bulging'. To some extent this might have deflected criticism from the  'ultra-purist'  dry-fly followers of Halford.

Commercial Tup's Nymph

But then the mystery begins. The next section is a bit pedantic, but I've included it in an effort to clarify the shape, colour, and contrast between the elements in Skues' version of the Tup's Indispensable nymph; the underlined factors may contribute to the nymph's  'prey image'.

Skues tied the "body rather fat". The Tup's Indispensables I have come across have a distinct fat 'thorax' - as illustrated here - and a relatively long body to the rear. It's worth emphasising that Skues used "three or four turns of "coarser untwisted primrose sewing silk at the tail,".

Courtney Williams (1973 p320) - who was in correspondence with Skues - writes: "Mr Skues's pattern in which he takes the dubbing mixture nearly the whole length of the body, only employing two or three turns of tying silk at the tail." [emphasis added]. This is consistent with the inventor's comment; Austin wrote to Skues: "I do not call the yellow at the end of the Tup, a tag. It is hardly big enough for that." Courtney Williams (1973 p318). There is a difference between the thickness of tying silk and untwisted sewing silk that will affect the final shape of the nymph.

Skues on colour

Tup's Indispensable dubbing made by Mr. Austen (sic), the originator. Larger image here

Austin's dressing was kept a secret by Skues to protect the Austin's family business. This led to commercial fly tiers trying to reproduce Austin's fly which led to ".. some most extraordinary patterns masquerade under the Tups marque. They range from the quite unbelievable to the truly impossible." Courtney Williams (1973 p318).

It is noticeable that in 1921 when Skues' most famous book The Way of a Trout with the Fly was published there is hardly any reference to Tup's Indispensable, or any reference to the creamy pink dubbing he described in his first book (Skues 1914 p 17).

I don't think Skues abandoned his appreciation of the importance of colour - especially red - in trout flies. For example in The Way of a Trout with the Fly  (Skues 1921) he commented:

"in flies purporting or intended to imitate natural insects, size and colour are the matters of consequence, ... shape is of very secondary consequence" (Skues 1921 p31-2) [emphasis added]

"It is, I think, beyond dispute that trout are extremely sensitive to red and are greatly attracted by it. " (Skues 1921 p26)

"One knows from the attractiveness of the Red Tag and the Zulu that they are peculiarly sensitive to red." (Skues 1921 p29)

Skues makes this insightful comment in the light of what was known, and not known, about trout visual perception in the early years of the last century. (cf Bowmaker & Kunz 1987, Douglas & Hawryshyn 1990, Gerl & Morris 2008).

"And the right fly is that which the trout finds to be the right colour. It does not always seem the right colour to the angler, and so it may fairly be questioned whether the trout sees colour just as man sees it. This is a question which deserves to be pursued further, but this is not the place to pursue it." (Skues 1921 p28-9)

Eventually in 1934, Skues published full details of the mixture used to form the fly's thorax: "The essential part of this dubbing is the highly translucent wool from the indispensable part of a tup, thoroughly washed and cleansed of the natural oil of the animal. ... There was also in the original pattern an admixture of cream coloured seal’s fur and combings from a lemon yellow spaniel, and the desired dominating colour was obtained by working in a small admixture of red mohair... "  (Quote from Skues in Flyfishers Journal 1934 in Robson 1998 p73-4) [emphasis added].

"When wet the Tup’s wool becomes somehow illuminated throughout by the colour of the seal’s fur or mohair, and the entire effect of the body is extraordinarily filmy and insect-like."(op cit) This suggests that Skues believed that the colour red was responsible for the effectiveness of Tup's Indispensable.

Skues admitted to "a constitutional incapacity to leave well alone, developing a number of variations of the pattern, according as it was to be used as nymph, spinner, or pale watery dun.. "My variations were made by using differing shades of seal's fur, not only in the reds but orange, yellows, cream and a range of olives, and in all shades I found the pattern useful." (Overfield 1977 p138)

Carrot Fly tied to Skues pattern by Jim Nice

Overfield (1977 p28) describes Skues 'Carrot Fly' as a joke, and gives this description of its body: "In three joints, tapered and fat. Tail end, pale yellow wool; middle, hot-orange wool; shoulder end, grenish seal's fur."

I'm not convinced that Skues was joking. In a serious discussion, Skues admits that imitating the colour of a nymph is difficult: "A dip of a muslin net into a clump of river weed would produce a large variety of colours, from pale yellow to darkest olive, and even to carrot colour" [emphasis added] (Robson 1998 p85). Skues modelled his Carrot fly on the Half-stone, and attests to its effectiveness. (Robson 1998 p85). Skues may have presented the Carrot fly in a whimsical manner (see Robson 1998 p165-6 ), but often people with heretical views present them in a light-hearted way to avoid ridicule - and Skues had plenty of that to endure from followers of Halford.

It's not clear what natural insect the Tup's Indispensable nymph tied with red seal's fur represented. For example, Courtney Williams (1973 p320) suggests: "As a nymph, it is especially attractive since it bears a close resemblance to those which have the appearance of bleeding at the thorax." I think this may not be the whole story because "The major difference between insect blood and the blood of vertebrates, including humans, is that vertebrate blood contains red blood cells. ". Insects have hemolympha which is clear or tinged with yellow or green ( DeSalle 2001).

Skues on shape

Modern tyings of Skues' nymphs (Lawton 2005 p56)

Reading Skues (Skues 1921 p17), and seeing the correct tyings of Skues' nymphs in Lawton (2005 p56), gives the impression that Skues discounted the importance of shape: "Taking it by and large, the fact is indisputable that the shabbiest, roughest, most dilapidated, most broken-winged fly is as likely to kill as the newest and freshest of the fly- tier's confections — provided size and colour be right. What is "right" must be the subject of further discussion. Meanwhile, I think I have established this, that in appreciation of form and proportion and detail the sight sense of the trout is defective."  But a disheviled fly still has a shape, and Skues recognized their appeal to trout. The shape he describes in now associated with flies that are trapped on, or in their attempts to burst through, the surface film, and present an attractive prey image of vulnerability.

Skues on size

Skues comments on size. On chalk streams he advises that "there is seldom any occasion to increase the size of the artificial fly above the normal size of the natural. ". But on freestone rivers " The fuller the water the larger the fly " is a good general working rule, (Skues 1921 p31), but he admits that he cannot give a convincing explanation for this advice.

In my opinion, Skues' advice on the colour and shape of nymphs is still relevant today, and is worth further exploration in light of the scientific literature dealing with search images and super-stimuli. But before descending down that particular rabbit-hole, it's worth exploring how American's expanded Skues' work on nymph fishing.

Tup's Indispensable crosses the Atlantic

You might think that Skues' Tup's Indispensable nymph has been forgotten in America. For example, it doesn't appear to be included in Dave Hughes encyclopedic and weighty (4lb 12oz) Trout Flies: The Tier's Reference (1998).

But it makes an appearance in his later book Wet Flies (2015) [which incidentally is an outstanding book] where it has become Leisenring and Hidy's Tups nymph which is "tied in a style close to, and perhaps based on G.E.M. Skues thorax nymphs".

There was a important interchange of ideas between Skues and "Big Jim" Leisenring; "Leisenring was the counterpart to Skues in England, exploring and promoting wet flies during a time of dry-fly dominance in both countries" (Hidy, Lance 2018).

Tup's nymph tied by Leisenring (from Hidy, V, S. 2019)

Skues' Tup's Indispensable was one of Leisenring's favourite flies (Hughes, 2015 p 115). Based on Skues nymphs, Leisenring and Hidy developed  'flymphs' ; flies tied to represent the transition from nymph to adult i.e. emergence.

Skues had a profound influence on American fly-fishing; this is reflected in significant advances in the design of flies to fish at various sub-surface depths (McGee 2007). "Skues ideas were so revolutionary they have influenced the fly fishing world ever since. In fact, many of Skues nymphs resemble the Thorax Style Soft-Hackled Nymphs we still use today. They have the realistic tailing, abdomen and thorax segment definition, and most importantly soft-hackled collars." (McGee, undated p8) [emphasis added].

Dave Hughes (2015 p335 & 115) includes the Tup's nymph in his recommended short list of flymphs, and remarks "This was one of James Leisenring's favourite patterns, and over time it has become one of mine."

British rams will give a sigh of relief when they learn - in this video of Dave Hughes tying a Tup's Nymph - that provision of the 'Indispensable' element of the fly now rests with the American fox !

However one element, the red / pink colour in Skues Tup's Indispensable, has made little impact overseas. This may be due to the difficulty of creating the right colour mix that Skues used to create the dubbing for the thorax - Mr. Austin's "Magic Dubbing" - Kaspar (2017). For example, here is a discussion that started 10 years ago on the internet-based  Flymph Forum

Brian Flechsig Tying Tup's Indispensable

"Somewhere, at this moment, some nameless but responsible fly tyer of good conscience is fretting over the near impossibility of finding exactly the correct dubbing for the thorax of a Tups Indispensable dry fly or nymph. " (Wickstrom 2000). Recently I found him - Brian Flechsig of Mad River Outfitters ties the fly, and explains its history. The first problem is that this "Magic Dubbing" was concocted in 1900 by a barber in Tiverton, Brian solves this problem by using a mixture of Senyo's Laser Dub (Pink, Tan & Red) The second more important issue, is that animals may not perceive colours the way we do. The human visible spectrum is shown in the diagram below; fish see ultraviolet, we don't. And there are differences between men and women in their perception of colour, which explains why I'm not allowed to choose paint ! (Gerl & Morris 2008).

The spectral sensitivity of brown trout is from 320 nm to 750 nm. (Bowmaker & Kunz 1987). Young brown trout have cones that respond to ultra-violet, but  "UV light seems to have limited practical significance to the adult trout. "  (Randall 2014 p87). Humans cannot see ultra-violet.

Image from  YouTube video:  How Trout See Color - A Research Based Explanation

The peaks for green and blue are similar in brown trout and humans. But there is a potentially important difference in sensitivity to red. The diagram shows this difference between humans and trout in sensitivity (peak response) to red.

The peak human response to red is at a wavelength of 550 nm

The peak trout response to red is at a wavelength of 600 nm

"This means that where humans see a dark reddish color, the trout sees a much brighter color and in a lower visible light condition" (Trout Pro 2010).

But this alone does not allow firm conclusions about colour vision, or its role in their behaviour (Douglas & Hawryshyn 1990 p397).

Behavioural studies are a more reliable method of determining what colours an animal can see. In daylight, when presented against a greenish-blue background, rainbow trout preferred coloured food in the following order: blue, red, black, orange, brown, yellow, and lastly green. At low light intensities the order of preference was yellow, red, blue, and lastly black (Ginetz & Larkin 1973).

It's very unlikely that Tup's Indispensable, or any fly, will only elict a trout's rise if it has a specific wavelenth of colour in its dubbing . This attitude stems from a literal interpretation of dictums in  selectivity theory , such as: "Based on my personal experience, I find that for tough, selective trout on today's pressure waters, even the color of the thread and hook matter." (Supinski 2014 p 68).

---

---

Supernormal stimuli

This rest of this essay provides additional detail for the  "What is a prey image?"  section in the essay Why does a trout take your fly? which is a chapter in my web-book The Heuristic Trout

The YouTube video is a lecture by Deirdre Barrett, Ph.D. on "Supernormal Stimuli"

A supernormal stimulus or super-stimulus is an exaggerated version of a stimulus to which there is an existing response tendency, or any stimulus that elicits a response more strongly than the stimulus for which it evolved. (From supernormal stimulus entry in Wikipedia )

This diagram, from the original research , shows that bees are significantly more likely to visit flowers with UV reflecting crab spiders on them.

Transparent plastic foil covering on each flower and spider ruled out the possibility that olfactory cues were involved.

This result is important because it shows that - unlike experiments with artificial supernormal stimuli that don't exist in nature - adding a natural super-stimulus, makes the flower more attractive.

Fig 1 (Heiling et al, 2003) Proportion of bees that visit vacant daisies (yellow bars) and daisies occupied by spiders (blue bars) in the presence and absence of olfactory cues.

How do supernormal or super-stimuli relate to artificial trout flies? The idea of artificial trout flies as super-stimuli lies at the heart of contemporary views on selective trout : "The right fly is one that resembles the natural so closely that the fish seem to prefer it over the real thing.. (Swisher & Richards 2018).

In ethological terms, Swisher & Richards want an artificial fly to act like a super-stimulus, one that is preferred over the competition from the surrounding flotilla of naturals.

But Swisher & Richards cling to   Halford's  selection theory which demands precise imitation of the natural insect. George La Branche pointed out the logical flaw in Halford's precise imitation approach: "the chance of the artificial fly being selected from among the great number of naturals on the water is one to whatever the number may be."

The super-stimuli studied by ethologists consist of crude artificial 'models' with abstract features that bear little resemblance to natural creatures. Therefore it might be argued - because of our inability to identify the crucial feature(s) - that creating a super-stimulus trout fly is an impossible task.

But we have already encountered an example of a natural super-stimulus that exists in nature, but remains hidden in clear sight. The dark and light-coloured moths in Kettlewells research  on melanism in the peppered moth became super-stimuli when one simple feature was manipulated - their camouflage.

Insects that trout eat are camouflaged to protect them from predators.

The moths were moved, from their normal environment, to an environment in which they were conspicuous, and consequently more likely to be eaten by birds. In a novel environment the moths became: "an exaggerated version of a stimulus to which there is an existing response tendency" (i.e. a superstimulus).

Against a dark background a naturally light-coloured moth became a super-stimulus, and against a light background a naturally dark-coloured moth became a super-stimulus.

This encourages me to explore simple features in artificial trout flies that might classify them as super-stimuli, and increase their attractiveness to trout.

By definition, a super-stimulus is an exaggerated stimulus that is more effective than the real thing in eliciting a behavioural response. In a previous essay I suggested that the Gold-Ribbed Hare's Ear (GRHE) may be an example of a super-stimulus. The GRHE is an enigma- it doesn't (to our eyes) look like a natural insect eaten by trout. Yet it has been a popular artificial trout fly for over 100 years, and remains so in its many guises.

Still from Jack Perks YouTube video of Male 3 Spined Stickleback in Breeding Colours

In Spring male sticklebacks change colour, establish a territory and build a nest. They attack male sticklebacks that enter their territory, but court females and entice them to enter the nest to lay eggs.

Niko Tinbergen discovered that crude simple models of aspects of natural objects were very effective in eliciting animal behaviour. In fact, the models were often more effective than the natural object. For this reason he called them supernormal stimuli. (I prefer the more recent term super-stimuli.)

Crude models with a red underside induce aggressive behaviour in breeding male sticklebacks. In contrast, a model with a swollen 'belly' elicits male courtship behaviour.

This behaviour is frequently presented in ethology textbooks. But it is not always easy to replicate (Rowland & Sevenster 1985).

---

Constructing a supernormal stimulus

Lesser Black-backed gull chicks peck at the red spot  (gonys spot)  on their mother's bill to induce her to regurgitate food (Point B in Fig 3.2 Ross-Smith 2009).

Ethologists view this as a classic example of a sign stimulus (red bill) eliciting a fixed action pattern (pecking). But it's worth pointing out that in reality there is variation in the potential of the gonys spot to elicit pecking from a gull chick. The implication of this 'variation' will, I hope, become clearer when we consider how stimuli, and super-stimuli, influence behaviour.

There is considerable variation between gulls in the size and colour of these gonys spots. Spot colour is a function of pigments (carotenoid and melanin) in the gulls diet. Spots with a poorly defined boundary are also less intensely coloured ( Ross-Smith 2009 Fig 3.12 ). Spot colour and definition is a good indication  ('honest signal')  of a gull's physical condition (op cit p 122-3), and has an influence on chick's behaviour.

Chicks will peck at an artificial model of a gull's head consisting of a red spot against a yellow background. The effectiveness of these models is a function of spot size, and colour intensity: "Small spots and those without a sharply defined boundary were less effective at eliciting pecks. " ( Ross-Smith 2009 p108).

However it is possible to construct a model that is even more effective than a real head by using a red pencil with three white bars at the end. This is an example of a supernormal stimulus.

In this experiment the supernormal stimulus (the Stick) received about 25% more pecks from gull chicks than the natural head, a model of an adult head, or a model of the adult's bill (Tinbergen and Perdeck 1950, Ross-Smith 2009).

---

Peak shift: An explanation of supersensitivity?

In her PhD thesis, Dr Ross-Smith comments: "Despite the acclaim that Tinbergen and co-workers received for the discovery of the supernormal response in gull chicks, little subsequent work has concentrated on the mechanism governing it...The question therefore remains - why does the supernormal pecking response exist?"

Part of the answer to the mechanism question may lie in research by experimental psychologists working on 'discrimation learning'. Discrimination learning involves teaching an animal to distinguish between two stimuli (S+ and S-) by reinforcing responses to one stimulus (S+).

In this video  Michael Domjan,  Professor of Psychology at the University of Texas at Austin, describes the concept of stimulus control, and its measurement with stimulus generalization gradients.

This learning situation is similar to that of a trout developing, and using, a 'search image' (S+) to detect food, and ignore other objects (S-). These experiments revealed a surprising effect (peak shift) that challenges the belief that we need to precisely imitate natural insects to cope with so-called "educated selective trout". In their book  Selective Trout  Swisher & Richards state:

"The fly fisherman who knows what is hatching and has realistic imitations will consistently be more effective than the angler relying on trial-and-error methods.. The right fly is one that resembles the natural so closely that the fish seem to prefer it over the real thing.."

Pearce (1977 p110-111) describes the results of an experiment (by Hanson 1959) in which pigeons were :

rewarded with food for pecking a response key illuminated by green light of 550 nanometers (nm) (the positive stimulus S+)

not rewarded when illuminated by yellow light of 590 nm (the negative stimulus S-)

It's no surprise that, after a period of training, responding was stronger to S+ (550 nm) than S- (590 nm).

Test trials involved examining the pigeons response to different coloured lights ranging in wavelength from 480 nm to 620 nm.

But it is a surprise that the pigeons response to the test wavelength of 540 nm (the novel stimulus S') was higher than responding to S+ (550 nm), the previously reinforced wavelength. This is called the peak shift effect.

Maybe peak shift only applies to pigeons in Skinner boxes ? No, it is found across species: "Peak shift is taxonomically widespread: exhibited by birds; mammals, including humans; fish; and at least some arthropods " (Lynn 2010).

Is peak shift a robust and reliable effect? Yes.One of the most consistent findings in the study of learning is that once an animal has been trained to distinguish between a positive (rewarded) and a negative (unrewarded) stimulus, an extreme version of the positive stimulus will elicit a stronger response than the positive stimulus itself  (Rhodes 1996).

What are the implications of 'peak shift' for anglers, specifically for Swisher & Richards' 'Selectivity theory'? Peak shift suggests that trout will prefer an artificial fly that is not an exact imitation of natural fly.

What explains peak shift? "Peak shift has been almost invariably attributed to Kenneth Spence’s 1937 theory of overlapping gradients of excitation and inhibition." (Lynn 2010). For example, Hanson's pigeon experiment can be explained in terms of Spence's theory of stimulus generalization (Pearce 1997 p110-111)

This diagram - based on Spence's theory - is an interpretation of the relationship between S+, S- and the super-stimulus S' - (From Vidya (2018) after Fig 5.3 in Pearce (1997)

"Figure 2. Gradient of generalization and peak shift illustrated. S+ (red colour) and S- (orange colour) are two stimuli that an animal is trained to discriminate between by providing a reward when it chooses red and no reward when it chooses orange. Each stimulus has a gradient of generalization around it, represented by the two curves. The excitatory generalization gradient (around S+) is larger than the inhibitory generalization gradient (around S-) because of reward associated learning. S’ (dark red) is a new stimulus presented to the animal. The difference between the two gradients, shown as filled set brackets, results in the tendency to approach one stimulus versus another. The animal prefers S’ instead of S+ that it had earlier been rewarded for because the magnitude of difference between the gradients is greater at S’ than at S+." From Vidya (2018)after Fig 5.3 in Pearce (1997)

Pearce (1997) describes a limitation of Spence's theory in accounting for an aspect of peak shift - revealed by a transposition test. According to Spence's theory many stimuli to the left of S' will have less of an excitatory influence than S' itself. Despite attempts, there has been no success in demonstrating this reversal effect in animal learning by experimental psychologists in a laboratory setting (Pearce 1997 p 112). But Spence's theory may still be a useful way of understanding the observed effect of super-stimuli on animal behaviour under natural conditions.

Rhodes (1996) discusses an alternative explanation of peak shift - Adaption level. Both theories predict that an exaggerated version of the positive stimulus will elicit a stronger response than the positive stimulus that was reinforced during training .

For some researchers, the 'peak shift effect' is so similar to the super-stimulus effect that the underlying mechanisms are identical. They regard the two terms as synonymous, and argue that the use of separate terms is due to enduring tension between behaviourism and ethology over the Nature vs Nurture debate (Ghirlanda & Enquist, 2003). For example, "Discrimination learning may not provide a general account of supernormality" Staddon(1975).

In a footnote, Rhodes (1996) repeats a caveat expressed by Baerends & Kruijt (1973) that peak shift may not be responsible for supernormal preferences because in laboratory experiments there is a clearly defined negative stimulus (S-) that the animal is trained to avoid. But in a natural situation (for example egg retrievail) it's not clear that S- is present. She suggests that this could be tested by seeing if inexperienced birds show a supernormal preference. In a recent report, Outomuro et al, 2020) found that great tits first may need to associate the stimulus to the reward, after which there may occur a preference for larger prey traits

For some academics, peak shift is either a heritable trait (Nature), or a learning phenomenon (Nurture). For others, peak shift and super-stimuli are the same thing. I suspect that the answer may lie somewhere in the middle, and involves a combination of nature and nurture, as suggested by Outomuro et al's (2020) finding. Consequently I am interested in the observations of anglers, ethologists, and lab-bound experimental psychologists.

---

Learning in a laboratory & behaving in a natural environment

What is the mechanism responsible for the super-stimulus effect? One ethological explanation is captured in this theoretical example. Arak & Enquist (1993) describe a very simple hypothetical situation in nature consisting of two flowers (Flower A and Flower B) that have evolved petals to attract pollinating insects. Flower B has evolved an elongated petal. Flower B also has more nectar than Flower A. The insect has evolved to distinguish A from B on the basis of the presence or absence of an elongated petal.

Arak & Enquist (1993) present this diagram, and figure legend, to describe this situation. The diagram has a set of three lines (labelled i, ii, and iii) to describe how an insect might react to flowers with a larger than normal elongated petal that do not exist in nature, ethologists would call such a flower a supernormal or super-stimulus.

Line i (yellow) represents the visiting behaviour of an insect that only visits flowers with a petal that is elongated within a very restricted size range. It will not visit any flower that lies outside this size range. This line illustrates an important point. It represents the behaviour of an animal that has a very restricted range of stimuli that it reacts to. For example, if a  gull chick  behaved in this way, it would not peck to obtain food if its mother had consumed food with sufficient pigment to provide the elements required to support a large bright red gonys spot. And vice versa, it would not peck a mother's bill if she had an relatively poor diet before nesting. Both gull chicks would starve. Similarly, if a goose only retrieved eggs within a restricted size range, she would fail to raise offspring in seasons with poor or good feeding conditions in the pre-nesting period.

This has an implication for the design of trout flies. It is very unlikely that even the most selective trout have a search image that is restricted to the extent shown by Line i (yellow) in this diagram. You could get the impression, from some fly-fishing literature, that selective trout do behave in this way:

"Selectivity, therefore, is not a conscious decision for the trout; it is a trained response. Such training in animals can be very precise... It's the same for the selectively feeding trout. The fish cannot consciously overrule its training. Once locked onto size 18 Baetis emergers, the fish is unable to select anything else." Borger (1991).

Borger (1991) is suggesting that trout would not rise to size 17 (smaller) or 19 (larger) Baetis emergers. But, like all creatures,  insects show variation  in size, shape, colour etc. The design of artificial trout flies can accommodate this variation.

Borger's (1991) advice to anglers is based on an outdated view (typology) of the variability found in nature that can be traced back to ancient Greece, but is challenged by evolutionary biologists. "There are two fundamental ways to view the biological world: As sets of nearly identical essential types (typology) or as populations of heterogeneous individuals (populational thinking)".  Typology assumes that all individual insects in a species are identical, and evolution proceeds through mutation, rather than on pre-existing genetic variation (populational thinking) (Powell 2018).

Trays of insects in the   Natural History Museum  are one example of the variability within a species of Ephemeroptera.

Line ii represents an insect that will visit flowers with a single petal that is larger than those on the natural Flower A, especially to the petal length indicated by the red circle on line ii.

Line iii represents an insect that will visit flowers with a single petal that is larger than those on the natural Flower A. But there is no limit to the length of the single elongated petal that will be visited by this hypothetical insect.

The point that Arak & Enquist (1993) make is that all three strategies - represent by lines 1, ii, and iii - enable the insect to gather nectar. They describe the behaviour exhibited in lines ii & iii as 'hidden preferences'. There is no reason for evolution "to maintain one particular mechanism of recognition" (op cit p 208). This explains why gull chicks will peck at a  range of stimuli  - from super-stimuli that contain the features (gonys intensity, size and definition) associated with good physical condition, to the weaker gonys stimuli of mothers in poor physical condition.

My take-home messages are:

The responsiveness of the hypothetical insect represented by line i (yellow) resembles the behaviour of the 'educated selective trout' presented in some fly-fishing literature, i.e. a trout that will only rise to an artificial fly that is a precise imitation of a particular natural insect.

The responsiveness of an insect represented by line ii is the same as the pigeons' behaviour (described above) where response to the test wavelength of 540 nm (the novel stimulus S') was higher than responding to S+ (550 nm), the previously reinforced wavelength. This is the experimental psychologist's 'peak shift effect'. It is important to point out that attractiveness declined after the 'peak' was exceeded. In addition, this effect was found when the animal was reinforced with food for learning to discriminate between two stimuli (S+ and S-). This is fundamentally different to ethologists' field experiments where animals were responding, with a fixed action pattern, to novel stimuli that they had no prior experience of, and no opportunity to learn about.

The responsiveness of an insect represented by line iii is an example of Tinbergen's supernormal stimulus. Note how the line does not peak, and then decline, instead it rises steadily until it reaches an asymptote. This picture shows that a gull will pull a very large egg back into the nest in preference to an egg of natural size.

Obviously there will be limits (due to 'mouth gape' Sánchez-Hernández et al 2012) to the physical size of insects that can be consumed by trout. The importance of using the correct size of fly frequently crops up in the list of factors that should be considered when designing trout flies. How do you decide on a range of suitable sizes?

One suggestion is to refer for guidance to the literature on the size of prey consumed as a function of age. For example, this diagram shows: " Box plots of the age-related variation in prey size of Salmo trutta in the River Furelos (NW Spain) during summer. The solid line within each box represents the median, the bottom and top borders indicate the 25th and 75th percentiles, the notches represent the 95% confidence intervals" ( Fig 7 Sánchez-Hernández et al 2012). Here is an explanation of Notched Box Plots

In natural settings animals (e.g. gulls, pigeons, bees and trout) need to respond appropriately to variability in potential food. This contrasts with the behavioural rigidity imposed on so-called selective, educated trout: "Once locked onto size 18 Baetis emergers, the fish is unable to select anything else" Borger (1991).

The next section presents a mechanism - signal detection - that accounts for how animals respond to the double-challenge of both identifying, and distinguishing between food, and non-food signals, with variable characteristics. .

---

A signal detection theory of peak shift

A recent study by Lynn (2005, 2010) is useful because it has elements of both the ethological and behaviourist approaches to studying animal behaviour. The subjects were groups of 10 bumble bees studied under laboratory conditions for their reaction to artificial flowers of different colour. The experiment involved bees that were trained to forage for nectar on flowers of a particular colour (a behaviourist approach), and others that received no training (an ethological approach), to see which colour each group preferred. In simple terms, this is a study of nature, nurture, and the interaction between nature and nurture.

Results of a peak shift experiment (Figure 1 Lynn 2010). These data are from bumble bees trained to discriminate between colors to forage on artificial flowers of different colour (Lynn 2005). Standard error of n=10 bees per group is shown. During training, the positive stimulus S+ was an artificial flower with a drop of sugar water; the negative stimulus S- was an artificial flower with a drop of saltwater.

A control group (black line) was trained to approach the S+ stimulus

A discrimination group (blue line) was trained to approach S+ and avoid S-

A Naıve group (red line) received no color training prior to testing

Then all groups received a test where the full colour range of stimuli was presented. Deionised water was placed on all the artificial flowers.
Results:

The Naıve group (red line) exhibited an innate preference for bluish flowers

The Control group shows that their peak response was slightly shifted toward the innate preference rather than centered over S+ itself

The Discrimination group (blue line) exhibited a peak shift in their preferred stimulus away from S+ in a direction away from S-. It is reasonable to conclude that the direction of this peak shift may was due to the impact of S- on their behaviour. I think the behaviour of this group is of more interest to experimental psychologists (behaviourists) than ethologists. [ I'd be asking questions about the equivalence in the relative reinforcement strengths of the two stimuli S+ and S- ]

It is interesting that the Control group (black line) showed an interaction between learning (nurture) and an innate preference (nurture) that showed a balanced outcome. In other words, the bees were able to adapt to what the environment provided (S+) during training, without abandoning their innate endowment which was expressed as a peak shift, or preference for a super-stimulus, during the test.

Lynn (2010) provides an explanation of peak shift based on signal detection theory .

The S+ and S- signal distributions (shown here as yellow and blue bell-shaped gradients) represent the subject’s estimate of the likelihood that a particular stimulus is from the S+ or S- stimulus class. Overlapping distributions produce uncertainty about which response (e.g., approach or avoid) is appropriate to give to any given stimulus. Behavioral response is dictated by a utility function (Box 1, eqn [2]) that integrates the signal distributions, the estimated probability of encountering an exemplar of either stimulus class, and the payoffs expected from correct and incorrect responses. The maximum and minimum of the utility function exhibit peak shift. (Figure 2 Lynn 2010)

---

Signal detection & trout behaviour

Signal detection theory was introduced in a previous essay in The Heuistic Trout.

There are two factors that influence behaviour on a signal detection task: sensitivity and bias. Bias and sensitivity are independent of each other (Makowski 2018a).

Sensitivity refers to the ease of distinguishing between two stimuli, in this case natural and artificial flies. This is represented by the amount of overlap between the two distribution of features of the natural and artificial. It is important to distinguish that from what anglers think about the similarity of artificial and natural flies. What matters are the features trout use to judge the amount of overap.

Bias is the extent to which one response is more probable than another.

For example, if the angler makes a clumsy cast or is visible, the fish may not rise to any fly, natural or artificial, because the response to predators (fleeing) takes over from the feeding response.

In contrast, if there is hatch of fly, or an increase in  invertebrate / behavioural drift  a trout will be more likely to rise in preference to conserving energy by resting.

Sensitivity and bias are calculated by allocating a trout's responses into one of four categories: Hit, Miss, Correct Rejection or False Alarm (Makowski 2018b).

This table uses these categories to classify the decisions of trout faced with natural and artificial flies.

Decision is:
Item is:	Rise	Don't rise
Natural fly	Natural eaten [hit]	Natural missed [false alarm]
Artificial fly	Artificial taken [miss]	Artificial rejected [correct rejection]

I find signal detection theory a useful way to think about trout behaviour; the fly-fishing terms 'selectivity' and 'presentation' are similar to the terms 'sensitivity' and 'bias' used in signal detection theory. A clumsy cast will bias a trout against rising. I try to use artificial flies that have features that overlap with the search image used by trout to select edible items on my local freestone rivers.

---

Super-stimuli & the Recognition Heuristic

Peak shift and super-stimuli have profound implications for the design of artificial flies that seek to present an attractive  prey image  to trout. The most effective trout flies contain an exaggerated feature(s) rather than trying to precisely imitate the typical or average characteristics of particular insects. Therefore, if trout are feeding selectively it makes sense to target them with super-stimuli rather than precise imitations.

Exaggerated signals are highly effective for a variety of recognition systems. When humans, other animals, and connectionist networks  [computer-based AI programs]  must discriminate stimuli, or categories of stimuli, performance is often facilitated by exaggerated traits or features that distinguish the alternatives.  (Rhodes (1996)

Super-stimuli trigger recognition.  This preference for extremes seems to be a fundamental feature of recognition systems (Rhodes (1996)

Gerd Gigerenzer describes the Recognition Heuristic

In fast-moving uncertain situations, such as uneven flow in a freestone river, less knowledge is better than more knowledge for making fast and accurate decisions. This is an example of using a recognition heuristic. (Goldstein & Gigerenzer 2002, Gigerenzer and Goldstein 2011).

An exaggerated super-stimulus that stands out is more likely to trigger a correct response (a hit in  signal detection  terms), than a miss or false alarm.

---

The properties of a super-stimulus

The first thing to bear in mind is that a so-called super-stimulus is recognizable as belonging to the same class of objects as a 'normal' stimulus. For example, a person who is 6 feet 6 inches in height is rare, but recognized as a person. Their height lies within the bell-shaped distribution of height. There are other things about humans that are recognizable, but often at a sub-conscious level. For example, human facial symmetry is relatively rare but recognizable. Facial symmetry is the human equivalent of the gull's gonys spot: a clearly defined large red spot that is an 'honest signal' of healthy physical condition. The majority of humans exhibit some degree of facial asymmetry.

This diagram shows the position of super-stimuli on a distribution curve of features that make up a trout's search image. The red area in this diagram represents the area occupied by super-stimuli. It is clearly within the S+ (search image) distribution, and does not overlap with the distribution of any S- (non-search image) found in nature. In other words, super-stimuli cannot be confused with other objects in the animal's environment.

The  research on gulls  revealed that the size, colour and shape (edge) of the spot on the mother's bill are features that vary between individual gulls. Exagerated versions of these features were incorporated into a model to create a super-stimulus (Ross-Smith 2009). It elicited pecking, so it was recognized by the chick.

When designing trout flies it may help to keep in mind that a super-stimulus is a simple model. It doesn't need to look real to us, but it must be recognised by a trout's search image (S+) . An introduction to designing dry flies as prey images with super-stimulus properties is  available here.

Conventional trout flies resemble the blue distribution labelled S- in the diagram. Their features overlap to a greater, or lesser, extent with those of the natural. Selectivity theory (discussed at length in this essay) encourages precise (Halford 1913) - or very close (Swisher & Richards 2018)- representation of the natural fly that trout are seen taking. It strives to create flies that overlap the bell-shaped distribution of features in the natural insect.

In contrast, some very effective trout flies bear little if any resemblance to natural insects. The question arises: Why are they effective? It's possible that these flies are super-stimuli. They correspond to the red area in the diagram - the tail of the distribution of features in the natural fly's search image. The area described by ethological theorists Arak & Enquist (1993) as 'hidden preferences'; features such as facial symmetry in humans, large bright red gonys spots in gulls, not often seen in nature, but they are there nevertheless.

Tinbergen and Ross-Smith (2009) showed that super-stimuli do not need to be presented in a particular "context". This chick is pecking a super-stimulus. The normal context of the mother's head and bill is absent.

This analysis has led me to suspect that the  Gold-Ribbed Hare's Ear  (GRHE), and - after dark - a  wake fly  may be super-stimuli.

Search images can be thought of as a distribution of features, for example sizes, shapes, colours and movements. Whether or not all of these characteristics need to be present to satisfy a trout's search image, or create a super-stimulus, is considered in the next section.

---

Heterogeneous summation

Heterogeneous summation is an ethological term that refers to how shape, colour, texture, and size of objects combine to release fixed action patterns. The rule of heterogeneous summation holds that these independent and heterogeneous features of an object are additive in their effects on behaviour.

The experiment described here is an example of creating an artificial 'supernormal stimulus' by modifying the size, colour and pattern on an object - in this case a bird's egg.

Herring gulls lay eggs in a shallow nest on the ground. The eggs are green or brown in colour and are covered in dark blotches (O'Hanlon et al. 2020). If an egg rolls out of the nest it is retrieved by the parent.

In an elegant series of experiments, Baerends & Kruijt investigated what properties of the egg signalled to the gull that it should be retrieved back into the nest. They removed two eggs from the nest and placed two dummies on the nest rim. The dummies were of various sizes, shapes and colours.

The investigators were interested in exploring the birds' retrieval preferences:

• Are bigger eggs more attractive than smaller eggs?
• What is the effect of egg colour on this behaviour?

The influence of egg size and colour on retrieval

Several types of dummy eggs were used:

brown unspeckled

brown speckled (i.e. natural coloured)

green unspeckled

green speckled

The simplified results of a large number of  titration tests  are shown in this diagram. The picture shows the influence of egg size and colour on retrieval. The numbers 4 to 16 indicate the relative sizes of the dummy eggs used in tests.

Results of the titration tests.

Baerends & Kruijt found that herring gulls:

prefer the larger of two eggs of the same colour

prefer the speckled egg over an unspeckled egg of the same colour

prefer natural coloured (brown speckled) eggs over brown unspeckled eggs

prefer green speckled eggs over green unspeckled eggs

prefer green eggs over brown eggs

Note that the gulls prefer a green background colour to the normal speckled brown colour of gull eggs.
For example, a size 5 green egg is more attractive than a size 6 natural coloured egg. Because they also prefer eggs with speckles, this leads to the surprising observation that a size 5 green speckled egg is more attractive than a size 8 natural coloured egg.

The important finding is that the preference for larger eggs remains the same when other features like speckling and colour are changed. This means that each feature (size, colour and speckling) adds a specific contribution that is independent of the contribution of the other features. In other words the features are additive in their effect on the gulls' behaviour. This finding substantiated the concept of heterogeneous summation, which holds that the independent and heterogeneous features of a stimulus situation are additive in their effects on behaviour.

O'Hanlon et al (2016, 2020) found that egg size is a function of the quality of the gulls feeding prior to the start of egg‐laying. Less maculated eggs (fewer and lighter spots) were found in colonies with a poorer diet. The authors commented that " Larger eggs can more probably result in surviving offspring (reviewed by Krist 2011), and egg pigmentation can play an important role in the successful development of the embryo (Maurer et al. 2011, Lahti & Ardia 2016)." It could be argued that egg size and maculation are 'honest signals'. "Honest signaling conveys information that is a true indicator of the underlying quality of the sender and it is useful to the receiver" Petak I. (2019).

---

Experimental Method - The Titration Technique

I've included this section on methodology for the sake of completeness, and because I admire its elegance. It can be skipped.

It proved impossible to carry out a simple preference test because when two identical sized dummies were placed on the nest rim, the gull usually revealed a marked position preference, retrieving the dummy on the left first (or vice versa).

To overcome this problem, Baerends & Kruijt employed a very elegant titration technique in which position was titrated againt size. In anthropomorphic terms they said to the gull

"OK, we know you prefer the egg on your left / right hand side, but if we put this bigger egg on the right / left hand side will you retrieve it first?"

This diagram shows the titration method of determining the value of an egg dummy.

The large brown circle represents the nest with one egg in the nest bowl and two dummies on the rim.

The numbers 7,8,9,10,11,12 refer to the size of the dummies, r is the ratio of the sizes of the dummies on the nest rim. The dummy chosen on each trial is indicated in black.

X is the model to be measured.

Results:

Row 1: Determination of the value of the gull's position preference. Starting from the left hand side, the first test with two size 8 eggs, shows that the right-hand egg is preferred. This shows that the bird has a preference for eggs on the right hand side. The next test (to the right) shows that this preference remains when a smaller egg (size 7) is substituted on the right hand side. The sequence of tests shows that the value of the position preference lies between r=1.3 and r=1.5.

Row 2: Determination of the value of model X. The control tests show that the position preference remains unchanged- the bird continues to show a right hand side position preference. The experimental tests show that the value of X lies between egg sizes 8 and 10. (After Baerends and Kruijt, 1973)

---

Neuroaesthetics and the Art of Fly Tying

Ethology has been described as "interviewing an animal in its own language" (Sapolsky) , an approach echoed in book titles by Datus Proper "What the Trout Said", and Bob Wyatt "What Trout Want".

Proper (1993) stressed the importance of designing trout flies; Wyatt (2013) used sign stimuli to create prey images.

Neuroaesthetics is the scientific study of how the human brain designs, creates and appreciates works of art. Fly tying has been called an art; the design and creation are products of a human brain, but the audience are fish, and their reaction is sometimes inscrutible.

On the other hand, it is remarkable how often the results are successful. It turns out that the cognitive processes humans use to design and create flies are congruent with those used by our critical silent audience.

Neuroaesthetics lecture by Ramachandran

The neuroaesthetic analysis of the arts (Ramachandran & Hirstein 1999) is built on principles that I also use in The Heuristic Trout:

the peak shift effect, and  super-stimuli

selective attention

heterogeneous summation

"As the physiologist Zeki (1998) has eloquently noted, it may not be a coincidence that the ability of the artist to abstract the ‘essential features’ of an image and discard redundant information is essentially identical to what the visual areas themselves have evolved to do. " (Ramachandran & Hirstein 1999).

That also applies to the process of designing a trout fly as a  prey image  that fits a trout's search image. Ramachandran (2010 at minute 32) makes the distinction between an image captured by a camera, and one created by an artist.

Designing a prey image resembles an artist creating a cartoon.

Ramachandran & Hirstein (1999) give this insight into the process, and is another reason for abandoning attempts to precisely imitate natural flies: "Consider the way in which a skilled cartoonist produces a caricature of a famous face, say Nixon’s. What he does (unconsciously) is to take the average of all faces, subtract the average from Nixon’s face (to get the difference between Nixon’s face and all others) and then amplify the differences to produce a caricature. The final result, of course, is a drawing that is even more Nixon-like than the original. "

Maybe Skues understood this point when drawing a distinction between imitation, representation and suggestion in designing nymphs: "The imitation may be Impressionistic, Cubist, Futurist, Post-Impressionistic, Pre-Raphaelite, or caricature. The commonest is caricature. It therefore catches most fish." ( Skues 1914 p78 )

A cartoonist works to grab the viewer's attention:"The third important principle  [ of neuroaesthetics ]  is the need to isolate a single visual modality before you amplify the signal in that modality. For instance, this is why an outline drawing or sketch is more effective as ‘art’ than a full colour photograph. "

Neuroaesthetics uses the concepts perceptual grouping, binding and contrast to explain how objects stand out from the background ( Ramachandran & Hirstein 1999, Gooch 2002). Perceptual grouping has its roots in Gestalt psychology ("the whole is greater than the sum of the parts"). It is similar, but not identical to the etholgical concept  heterogeneous summation  ("the whole is equal to the sum of the parts").

Grouping, binding and contrast allow a predator to detect   camouflaged cryptic prey.

Selective attention  enables trout to use a  Recognition Heuristic  to detect prey.

A fly-tier can be compared to an artist. In his talk   Ramachandran  makes the point that the artists we admire, and whose work has survived through the centuries, have all unconsciously tapped into basic perceptual processes in the brain such as peak shift and supernormal stimuli. The same can be said of some artificial flies used to catch trout. They consist of one or more sign stimuli that fuse to create a prey image that matches the trout's search image.

There is an obvious, but often overlooked, problem faced by those designing trout flies. Jacob von Uexküll pointed out that each animal species lives within a sensory world created by the capabilities and limitations of their particular sensory systems (Burghagen & Ewert 2016). For example, a dog's sense of smell, and ability to track criminals or detect illegal objects, surpasses our odour perception. von Uexküll coined the term 'search image' to describe this ability. Dog handlers train dogs using olfactory 'prey images'. But dog handlers have one important advantage. They may not be able to smell the prey image themselves, but they know the source of the smell - money, firearms, narcotics, and can put that scent onto a training dummy.

Compared to a fly-tyer, the dog-handler has it easy. They use reinforcement to train a dog to perform a behaviour. The fly-tyer is trying to trick a trout into performing a response for no reinforcement. It gets worse. When we design a trout fly as a prey image, we only have a hazy understanding of what constitutes a trout's search image: "A search-image reproduces the original less from the complete image of the object being sought, but rather focuses attention on certain cues of the search object." (Burghagen & Ewert 2016). But we do have a much clearer appreciation of what 'prey image' will catch a fly-fisher - the market shapes the product design. That's why 99% of trout flies are a marketing miracles. They are designed with the wrong prey in mind !

---

Movement - the Overlooked Super-Stimulus?

It could be argued that there has been excessive concentration in the fly-fishing literature on the size, shape and colour of artificial flies. This is understandable because they are relatively easy to control during the construction of an artificial fly.

Wyatt (2013 p 97) adds 'behaviour' to the essential features of a  prey image. "What the trout responds to first are the most obvious aspects of the prey: size, shape and behaviour. If one of these essentials is missing or the fly behaves unnaturally, the lifeless artificial doesn't send the necessary "eat me " signals, the trout's go button isn't pushed, and it lets the fly go as "not food" ".

I want to focus on one aspect of behaviour - active movement of the prey image. A previous essay described an experiment by Bianco et al (2011) showing that moving spots which stood out from the background evoke eye movements and J-turns of the tail, which are defining features of natural hunting in larval zebrafish.

These results have been replicated, which supports the conclusion that zebrafish hunting responses are triggered by a combination of visual features. "Size, contrast polarity [a bright spot on a dark background, or a dark spot on a bright background] and speed of motion interact, such that stimuli that are large, dark, and fast are most effective in triggering hunting responses." (Bianco & Engert 2015).

Bianco & Engert (2015) discovered direction-selective brain cells (in the rostral tectum) that responded to large, dark, moving spots - i.e. the most effective stimuli combination for eliciting hunting behaviour. This effect of a combination of sign stimuli triggering a fixed action pattern is an example of heterogeneous summation.

Detailed results that support this conclusion are available here 

Lagogiannis et al (2020) found that experience of feeding larval zebrafish on live prey increased capture efficiency. The authors discuss the possibility that refinement of heuristics, for example the recognition heuristic, and the tracking heuristic  are responsible for development of improved hunting efficiency:

" We cannot exclude the possibility that larvae posses an  adaptive toolbox  (see Todd and Gigerenzer, 2012; Todd and Gigerenzer, 2007) of distinct, preset, hunting strategies. This would imply that the role of learning is to utilize experience to find the best match between the set of available hunting behaviors and the particular foraging environments a larva encounters." Lagogiannis et al. (2020) [emphasis & link added]

---

Is speed of movement a super-stimulus?

The application of the Tracking Heuristic to explain how a trout rises to intercept a fly on, or trapped in, the surface film is covered in depth in a previous essay  "How does a trout catch a fly". In that situation, fly-fishers go to great lengths to minimise movement of the fly to avoid drag.

In contrast, movement can be a positive advantage when presenting sub-surface flies. It is possible that several techniques that induce movement convert a life-less artificial into a super-stimulus.

Frank Sawyer's 'Induced Take', 'Leisenring's lift' and 'Hidy's subsurface swing' as   described here, are ways of introducing movement into the behaviour of an artificial fly.

In Sawyer's induced take the fly is allowed to sink, and drift downstream towards the fish. It is then deliberately lifted towards the surface in a movement known as the “induced take”

Leisenring relied on stopping the downstream drift, and allowing the current to lift the fly in front of the trout as the line came to the end of its downstream swing.

When fishing the wet fly downstream, let the fly dangle for a few seconds at the end of the cast because trout often take a wet fly 'on the dangle'. This pause will have the same effect as Sawyer and Leisenring's methods; mimicking the behaviour of an insect ascending prior to hatching.

I found this a most telling comment from experienced West Country fly-fishing instructor David Pilkington (1983) : "Beginners should start off with the downstream wet fly because it gives the indifferent caster a chance of a few fish and helps to boost confidence". In other words, let's not make this over-complicated for beginners or the trout! Attitudes to fishing the wet fly down-and-across are discussed here

It is not clear from the scientific literature, what impact if any, the speed of movement has on the attractiveness of the fly. For example, I have not found any research on movement and peak shift. This isn't surprising because peak shift is a laboratory-based technique carried out in Skinner boxes with static or fixed stimuli that do not move. For example,  peak shift  has been demonstrated for a wide range of stimuli dimensions - wavelength, size, line angle, auditory frequency or intensity - that can be presented in a Skinner box (Rhodes 1996).

But it is clear from Bianco & Engert's (2015) research that movement is an important trigger for hunting in larval zebrafish, and the  toad  discriminates between prey and non-prey on the basis of size and movement.

It is interesting that effective wet-fly presentation techniques involve moving the artificial fly upwards towards the surface, and thereby reducing the distance between the fly and its reflection in the mirror.

---

Is symmetrical movement a super-stimulus?

"Symmetry is a key feature observed in nature (from flowers and leaves, to butterflies and birds) and in man-made objects (from paintings and sculptures, to manufactured objects and architectural design) " (De Luca et al 2019).

Animals can learn to use symmetric displays, and come to regard such displays as more attractive than asymmetric ones. For example, Kral (2016) reviewed evidence from field studies that bees found perfectly symmetrical artificial models of flowers more attractive than less symmetrical natural flowers, i.e. the artificial models were super-stimuli.

Recognition of symmetry enables birds to detect food (Mascalzoni et al 2012). Two day old chicks that had eaten preferred to peck at symmetric than asymmetric visual patterns. Clara et al 2007 concluded that: "The emergence of a preference for symmetry seems to depend on visual/behavioral experience obtained from pecking at food objects, independently of the kind of food (i.e., its shape)"

There are two theoretical explanations for the importance of symmetry:

Symmetry preferences evolved to identify high-quality food or mates. There is a growing literature on this possibility, a large part of which is devoted to human mate choice.

Symmetry enables 'figure-ground segmentation'. Figure-ground segmentation refers to the process by which the visual system organizes a visual scene to enable animals (and humans) to distinguish objects from their backgrounds. This is a fundamental process in distinguishing prey from its background.

It seems reasonable to suggest that symmetrical movement may be a super-stimulus because:

1. The most effective stimuli combination for eliciting hunting behaviour in larval zebrafish are large, dark, moving spots that activate direction-selective brain cells in the rostral tectum (Bianco & Engert 2015).
2. The up-and-down movements of natural and artificial nymphs create horizontal (reflective) symmetry in the trout's mirror (De Luca et al 2019).
3. Angler-created movements of sub-surface artificial flies are an effective fly-fishing technique (Pilkington 1983, Hughes 2015)

---