Thread by @evolvingcalm

@normonics' Intro to Applied Complexity #ACS101 #SpringA2021 Highlights

Session 12: Perception, Action, and Mind

(Back after a bit of a hiatus - have still been chipping away little by little in the background all the while)

Thread/

https://twitter.com/evolvingcalm/status/1369161036457111554

On:

How units, especially complex multicellular animals, mainly get around in the world in a macroscopic way.

Issues to do with perception/action and in some sense mind.

1/n

Gigerenzer, Gut Feelings -"Many skills lack descriptive language"
JJ Gibson -"The environment isn't the same as the physical world"
Maturana etc. -"The [eye's] operations thus have much more the flavour of perception than sensation, if that distinction has any meaning now"

2/n

People make decisions they don't necessarily know why.

e.g. gut feelings, intuitive decision making, unconscious processes.

(Compare & contrast that kind of decision-making with very explicit computation/calculated types of decisions)

3/n

In many, many, many situations people perform better when using their intuition - using their unconscious. Often when you try something that looks sophisticated e.g. writing down pros/cons - calculating this & that to make a decision - your performance is actually degraded.

4/n

This is contra a lot of cognitive science decision-making literature which really promotes using these tools and calculating odds and things like that.

5/n

Not only do skills lack descriptive language but what we think we do and what we actually do are often at odds with one another.

6/n

The paper 'What the frog's eye tells the frog's brain' is 50 years old now - very clear with respect to this issue of sensation & perception - rebukes some of the (namely) philosophical dichotomy implicitly assumed in the sense between sensation & perception - meaningless.

7/n

Perception starts in the case of visual perception as soon as the light is hitting the retina.

There's no sensation followed later by perception which is commonly how things are talked about still.

It's been 50 years & people are still sloppy around issues like this.

8/n

We move around the world not just because we happen to - it allows us to deal with the world's complexity, maintain access to things we need/resources, avoid danger/harm, create a better situation for ourselves etc.
To behave in this way demands the ability to perceive & act.
9/n

The 'ecology' of visual perception is about an organism situated in its full environmental context, in its full complexity, in its full richness.

As opposed to a setting: isolated, contrived, controlled.

What is visual perception to the organism living its life.

10/n

1 way organisms achieve viability in a fluctuating environment, most notably (& not only) animals, is by acting on & in the environment - moving around & interacting.
Perceptions are really 2 parts of what in essence is an inseparable whole - engagement with the environment.
11/n

The 'parts' of perception are named which give the impression of an input/output system - which in a lot of ways they really aren't, & animals/organisms are really not structured that way.

We are not 'input-process-output'.

12/n

Perception is a continuous cycle where perception-action are always coevolving together in time.

13/n

The mind emerges out of perception, action, environment.

Mind isn't necessarily just an emergent property, just of the brain, but of this relationship between perception, action, and that relationship with the environment that happens within.

14/n

Most attempts to treat mind in an abstract isolated fashion as in just the brain suffer from a scoping problem.

(You need sufficient scope to observe emergent properties.)

How big does the scope need to be to capture the emergent properties of the thing called mind?

15/n

The organism/environment pair/system comes as a whole together.

Perception is neither subjective/objective but relational or an emergent property of the system.

16/n

This understanding gets us around a lot of philosophical pseudo-paradoxes.

It's about the very things fitting together & interacting with one another.

17/n

Behaviour is not facilitated by a central controller.
It's tempting to think that way sometimes because we have something we call a 'central nervous system' & indeed most of the nervous system is bundled up into the skull & some of it is distributed it out.

18/n

It turns that the way we actually perceive, the way we use cognitive functions, we're often leveraging part of the body & part of the world that we're interacting with as part of these processes.

19/n

It's a scoping thing - you can't really account for the behaviour of a system in total if you try to isolate a part of it like just the brain or just a part of the brain and treat the behaviour of the system that way - you'll always be leaving something essential out.

20/n

Behaviour is a decentralised and distributed process that involves brain, body and interaction with the world.

21/n

Bernard Scott - "An organism does not receive 'information' as something transmitted to it, rather, as a circularly organised system it interprets perturbations as being informative".

(Another context with this concept of closure.)

22/n

Consider a system that is operationally kind of closed on itself & the environment is perturbing it e.g. photons in the retina are not 'transmitting' pre-packaged information but rather you have nerve cells that are being perturbed by those photons.

23/n

Environmental perturbations produce different kinds of behaviour due to those perturbations and systems use that somehow - leveraging it.

A subtle but important distinction.

24/n

An easy assumptions to make is the mind exists in the brain, or maybe more sophisticated: the mind is emerging out of the operations of the brain, that's a common assumption to come at this with & it's reasonable to make this assumption - maybe it's true, maybe it's not.
25/n

The brain is massively complex in possibility space.
Take 1000 connections per neuron - imagine each nerve on/off - that's 2^100 billion possible states - just enormous.
(But to put it in perspective we maybe aren't unique in an absolute sense - elephants are smart too.)
26/n

We never have really seen a brain in total isolation.

It's connected to the body - in some sense it comes from the body.

Ultimately it doesn't make too too much sense to study the brain once and for all in isolation but we need to study it in its context to understand it.

27/n

Maybe you can have mind without brain?

Bacterial chemotaxis - attracted to sugar crystals - sense the gradient of a sugar crystal & swim to it

Very complex systems & organisms that behave in the world - perceive and act on the world - don't have brains at all.

28/n

Maturana (RIP)/Varela would say yes - what makes mind is an autopoietic system that's acting in an environment/world.

Thus nervous systems incidentally are part of systems like that but not necessarily.

29/n

What is the scope of mind? is it just the kind of gooey stuff in the head? is it that combined with the body? or do you actually have to include all of that?

30/n

There's things you can learn about by studying the brain.

There's thing you can learn about by studying the brain and body.

And there's things you can learn about studying the whole system in the world

You learn different things depending on how you scope it.

31/n

There's this central control idea of a 'little man' sitting in the head a 'homunculus' & there's a 'theatre of consciousness' in philosophical discourse - a movie screen & the little man gets all the information & sends out all the instructions for the body to do.

32/n

Several problems with this idea - one is sheer complexity of behaviour we have a problem where we're trying to reduce the amount of complexity to a smaller system and have enough richness in that system to guide all the details that larger system needs.

33/n

Social systems, political systems, have the problem of central control & bandwidth constraints & how resolution constraints when you try to control a system centrally and the impacts of that/its implications.

34/n

Another problem - if there's a little man in the head driving the body well who's driving that little man? and so there's this 'infinite regress' - the little man must have a little man must have a little man etc. - something deeply problematic with that and to be avoided.

35/n

Hard to not get pretty philosophical when you start dealing with the mind.

One way Philosophers talk about what goes through the mind is 'mental representations' e.g. the mind represents things, often things in the world, maybe beliefs.

36/n

But often the assumption is really these representations are of objective physical properties of the world such that we can model the world in the head, represent it, act on the model, figure out what to do and then send the instructions out. Seems kind of fragile it is.

37/n

Although maybe it's kind of reasonable maybe what we're representing is not so much the objective world as it exists outside of us but something a little more related to how we're structured.

38/n

AI is a nice place for toying with some of these ideas.

People spend a lot of time and energy trying to build systems that do something.

39/n

There are roughly two camps historically involved in AI research:

symbolists vs connectionists

They have wax and waned in favour in different forms sometimes one is in vogue and then the other.

There are also hybrids.

40/n

'I recommend you use duckduckgo'

41/n

Symbolist conceptual systems are useful in that as a human observer you can look at them & understand what's going on, can relate to the ideas& concepts graphed.

Now back in vogue, not referred to as expert systems anymore but knowledge graphs.

42/n

Connectionists build artificial neural networks which have a graph structure where you have lots of nodes -just very simple internally they are pretty much 'neurons' that can be on or off & they are connected to one another typically in massive networks.

43/n

~Our understanding of the brain:

Node of cells called neurons fire action potential discrete signals down an axon which receives input from bunch of different neurons & integrates the firing - above some threshold it will fire & pass that down the line through the network.

44/n

But in a neural net the abstractions tend to strip things away. Things like cells coating neurons etc. probably do have relevance to how our brains works.

AI projects just treat artificial neurons as very simple on/off units that can be connected together, integrate input.

45/n

What an artificial neural network does is not to be found by looking at what any of its neurons does individually or in an isolated fashion but looking at the pattern the organisation of the system and how these patterns evolve overtime (generally organised to a useful end)

46/n

Symbolist systems are really bad at embodied tasks like robots acting in the world, very fragile to their assumptions but you can look & make sense of it.

Whereas in a neural network you can do things like pattern recognition but you can't really understand the details.

47/n

Two main types of artificial neural networks are 'feed forward' (unidirectional) and 'recurrent' (can loop).

The fundamental kind of feed forward network is called a perceptron.

The most basic kind of recurrent network is a Hopfield network.

48/n

With feed forward: one neuron each connection has a weight associated with it, how strong is that connection is often what you're trying to determine when you train a network.

With recurrent sets of neurons everything connected to everything. Threshold behaviour very common
49/n

The Hopfield network is like a kind of memory or perception. Think energy landscapes

It's a way of taking something that's gonna be different every time & mapping it to something that's known or common & that's why sometimes it's called content addressable memory.

50/n

‘I don’t want to spend too long on neural networks.

Always takes longer than I think it will.’

51/n

Visual Perception:

When we talk about perception we tend to talk a lot about visual perception because we are very visual creatures, so it's kind of handy.

Certainly we probably overemphasise it.

There are other modes of perception they offer different possibilities.

52/n

The retinal 'image'
So related to this distinction between sensation & perception that's often injected as a matter of axiom is this idea of a retinal image, with a world out there & rays of light come in, almost like a photograph - it's often assumed that's what's going on.
53/n

'What the Frog's eye tells the Frog's brain' idea questions this 'retinal image' idea
Why? Physiological evidence
They found a frog's retina doesn't behave like an image but actually responds to patterns - spatial/spatio-temporal ones - not to pixels, it's not a light bitmap
54/n

There's actually a very common pattern that's all over our retina & other animal's retina & it's something called a centre-surround pattern - looks familiar to a long range-activation local-inhibition pattern.
The difference being similarities to a feed-forward system
55/n

The receptive field is looking for patterns that are about so big, looking for contrast, looking for a spot.

It's a spot detector.

56/n

In our eyes we have a central fovea where there are these finer grained centre-surround patterns & in the periphery they are bigger, & your peripheral vision is blurrier so to speak & this is one of the reasons why.

57/n

We are doing a multi-scale band pass filter type operation with our retina, and so we’re looking at the patterns at different scales and we’re not looking at the pixelated image like you see in a photograph.

58/n

We are also sensitive not only to spatial patterns but the temporal aspects too. Immediately at the retina neurons are interacting - not only in this feed-forward way (actually an overly simplistic comparison to get the idea going) - they're also interacting laterally.

59/n

There's nerves returning from the brain into the retina & interacting. Right at the retina - where light begins to touch the nervous system there's an ongoing dynamic sensitive to spatiotemporal patterns but also actively modulated constantly by our own internal dynamics

60/n

We're never ever just getting a 'retina image' it's not a real thing, it doesn't exist, so the fact that it starts that way *there* at the retina already starts to hint how the whole system really is - one big irreducible system & it's not a kind of feed-forward type system.
61/n

Detection of motion can happen all the way down at your retina and indeed it is happening on the frog's retina.

It can see things before it gets to its brain

So in a sense it's preprocessing but maybe pre-what? It's completely interconnected.

62/n

If you take a neuroscience grad program they'll talk all day about the stages of processing but it's not really doing that - rather it's all recurrent on itself. We have this idea that there's stages in some sense (maybe true) but in another way there's nothing before/after.
63/n

There's one system that is dynamically evolving over time and there's no before or after it except in the sense of the clock's ticking something happens one moment then something happens the next.
But it's not like a procession along a factory line towards an outcome.
64/n

In terms of the vision and eye there's a constant churning of dynamics in the system and when the light comes in it just perturbs the dynamics.

65/n

There's a lot of philosophical confusion in the neurosciences - it's a minefield.

A lot of hang ups around models.

66/n

The dynamics are continuous whether your eyes are open or not. The part considered the visual cortex when you close your eyes it synchronises into about a 10Hz pattern and oscillates all together and then synchronises when you open again and starts behaving less coherently.

67/n

Brian shares his had his thumb detached as a kid then reattached without nerves

A Weird thing because phantom pains, even though it's been a number of years

'it's not there but it's still part of the system and it's strange.'

68/n

Joe talks about Ramachandran and the book 'Phantoms in the Brain' and tricking the brain.

Very strange stuff where very rapidly you claim ownership of things you perceive in synchrony with sort of multiple signals converging in tight correlation.

69/n

Gestalt psychology
A movement coming out of Germany where they were looking at visual perception as wholes and paying attention to the relationship between wholes and parts of images and how we perceive them.

70/n

See the upside-down triangle (negative space). Not there in some sense, there's no boundary mostly but you kind of perceive a boundary, reminiscent of like Hopfield network where there's some kind of attractor state in your brain & this is enough to perturb it into that.

71/n

Once you see it you can't unsee it.

72/n

Gestalt perception is something in AI they have a lot of trouble with, it's an interesting problem area.

73/n

Related is one shot learning where once you get it you get it you don't need more training or anything like that it is what it is. It's amazing how persistent these are.

74/n

You can look at things & you can't see anything at all then it clicks like a phase transition *click* there it is & then you can not look at it again for a year
(or a couple weeks can confirm)
& you can come back a year later & there it is.
The brain is amazing for sure.

75/n

Direction of constraints - so related to how we've been talking about things scientifically

There's a relationship to how we do perceive and how we've been talking about characterising systems in terms of constraints and possibility space.



76/n

When you see it in motion you can now understand what you're maybe seeing, but if you saw a stopped frame to begin with you probably wouldn't know what you're seeing.
When you see it in motion now you're learning the relationships & the constraints & the possibility space

77/n

Really how narrow it is for those dots, very little visual information but actually we naturally pick up on the constraints & how things are related to one another.

78/n

A great visceral example of what we mean by constraints in possibility space.

If these dots were all moving around independently you wouldn't see anything and they are moving around in a way that statespace is constrained and you see the system.

79/n

Closure of the nervous system
Again have this issue of what's going on here?
Ashby claims the nervous system is closed like very simple deterministic systems as the set & transformations always resolve on a state that's in the set that continues/transforms in that set etc.

80/n

The nervous system doesn't really care about how that closure is achieved - could be that a nerve is connected to another nerve with an axon sort of direct physical connection or it could be connected through the world.

81/n

For most people much easier to solve say a long division or hard multiplication by writing something down and that's connecting certain parts of the nervous system to other parts of the nervous system it just happens to go through the world in this case

82/n

We're able to achieve a lot more, it turns out, when we leverage that - there's no contradiction between this idea that there's something like closure in the nervous and there's other things that are not the nervous system that are helping to achieve that goal.

83/n

Sensorimotor contingencies:
The idea that when you move in space there's certain events that happen - sensory events let's call them & they are related to how you just moved.

84/n

For instance something called an optic flow pattern & whatever it's moving towards the entire optic array will be looking to flow outwards from that point - a central point where all light seems to be flowing outwards

85/n

Similarly if you're moving directly away from some point all of the optic information will seem into flow into a single, sort of sink, point.

86/n

if I stand in a room & the room is static & then I start walking forward I'll get optic flow & then I stop walking & then it stops so there's this relationship between what I'm doing & the feedback I'm getting from the world & that's this idea of sensorimotor contingencies.
87/n

Heuristics - in contrast to lots of internal processing to do things.
In (so many places e.g.) the 'Selfish Gene' Dawkins says something like 'oh yknow when someone catches a ball it must be there's something in their brain doing something like solving equations of motion -
88/n

- parabolic trajectory of motion to predict where the ball will be. There's a sort of model of physics in the head so they can catch the ball - they're actually solving in, if not unconscious sense, at least explicit sense, a model of physics in the head in order to catch.'

89/n

But in 'Gut feelings' Gigerenzer relays a finding that what people are actually do to catch a ball is just leverage a set of heuristics that are interrelated.

90/n

Roughly they use a heuristic where once the ball has reached it's maximum height & starts to drop again what they do is that they put their eyes on the ball & then they move in space forward/backward/left/right whatever they need to do in order to maintain the angle of gaze.
91/n

The angle is always constant so if the angle started to drop they would speed up, if it started to rise they would slow down or back up & once they lock it in & move in such a way that their angle of gaze never changes they are then on an intercept path with the ball.

92/n

The anecdote goes a new coach came into the team & some hot-shot outfielder was doing really good. The coach, trying to assert his dominance, said 'hey you're a hot shot but I see you slacking out there catching the ball just jogging so I want you to run fast, hustle.

93/n

And then the outfielder couldn't catch any balls anymore.

Because he didn't know how he was doing it either.

He couldn't say 'actually I use this heuristic, this gaze heuristic and you know I'm running the speed I need to run to intercept the ball.'

94/n

If we think about this as a system to understand how the player's catching the ball - we need to scope out to this whole interaction.

You can't find how does he catch the ball just inside the head it's not in there.

95/n

Sure the stuff inside his head is part of the system & it's necessary to do a lot of the things that are happening but the solution is not a set of equations running inside his head that model the trajectory of the ball - he's leveraging heuristics to get to the ball.

96/n

Indeed actually calculating the trajectory of the ball would be quite complicated with things like wind resistance & the air is moving, there's a breeze this & that,

This heuristic is robust to those perturbations as longs as he constrains his behaviour to the heuristic

97/n

The outfielder doesn't need to calculated those little deviations i.e. they're calculated implicitly by the way he moves - so the movement in the world is the computation - is the calculations.

'One' can also leverage heuristics in collective systems. E.g. Ants

98/n

Absorute shamefur dispray

アボソルテシアメフルデスペレユ

99/n

Compare

100/n

Honda put a lot of money into a robot that basically operates on the philosophy of 'model the world in my head and act on the model the send instructions to my body on how the model says I should act.

Essentially one of those symbolic systems.

101/n

vs Boston dynamics' Atlas - alternate way to engineer a system.

How'd they built it?

What they do differently - still has some symbols

They are using some neural networks they are also using heuristics.

102/n

When Atlas recovers, when he takes a misstep, there is a heuristic used.
When he starts to sense he's off-balance one thing that happens is one leg will be in the air.

So what happens? He actually relaxes the leg

& then the leg finds gravity.

103/n

The leg just falls downwards, now it's in a good position to catch him, no internal calculation
leverage gravity & just let the limb figure it out.

So it's not just the thing's brain thinking
the body's thinking too, so to speak.

Totally different approach to engineering.
104/n

Organism/Environment complementary
&
Affordances

'The environment for the organism is not the same thing as the physical world.'

105/n

Well what does Gibson mean by that?

The world that the organism, not just experiences but lives in, is the world of the actions that it can talk & the actions that it can take depend not only what's say physically in the world but the structure of the organism itself.

106/n

There's certain rocks out there that have certain densities & luminance's & all these physical properties to them - that's fine & good some of those rocks are smooth & steep & I can't grasp them, other ones are jagged & have cracks in them & I can grab them.

107/n

They're grabbable - they're not grabbable objectively.

There's nothing that's objective about grabbability.

But relative to my structure certain rock faces have grabbable pieces to them.

108/n

Essential opportunities for action.
The environment is this full set of affordances for the organism - what can I do in the world? & that is answered by, not only what's in the world, but what I am - how I'm structured - what skillset I have - what know-how I have.

109/n

A very good way to think about what is perception?

Perception is a set of affordances:
What I can do in the world & not subjective.
It depends on my actual structure & the world's actual structure & how they fit together or don't.

Not subjective or objective
Relational

110/n

Perception is not objective, the world without reference to me has no affordances, neither is it subjective.

It's a relational property.

It requires a certain amount of self-awareness in a very literal sense - it is a property that is a scoping thing.

111/n

Take a hand - tree & branch

Is the branch grabbable?

To ask that you have to reference to both the branch & the hand/agent that's doing the grabbing.

112/n

Perceptions are something that the agents are experiencing, but they're experiencing the relation between themselves & what's out there & what's out there is essentially provided structure & constraints on what they can & can't do but it's not determined.

113/n

The idea of affordances is really powerful way of thinking about a lot of what perception is and so we often see this in design settings.

114/n

Some bad examples:

"You've probably experienced this in a university if nowhere else."

115/n

You don't always need complex solutions to solve complex problems.

Sometimes simple solutions can solve complex problems.

116/n

On the surface appears at odds with the idea of requisite variety & need to meet it.
In fact there's no conflict at all, just some crosstalk going on with terminology.
Requisite variety is nearly tautological & thus essentially true in of itself.
Can't be negotiated with.
117/n

Looking at something like catching a fly ball it's not that the outfielder is using something simple to solve something complex he's actually just hitting the requisite variety exactly.

118/n

He needs this many degrees of freedom to solve that embodied problem - the problem of calculating the trajectory in an abstract sense is actually a different problem - he's having the exact right amount of complexity for the problem.

119/n

Moreover this is related to the likes of cell automata where we have 'simple rules' & they generate complex behaviour & there's no contradiction there.

The total system is complex & the behaviours it's generating are also complex.

120/n

Often complexity can be generated from iterating on things that locally are simple in appearance.

But the system isn't simple, the rule set is simple, so there's actually no contradiction there at all.

Achieve variety through a simple set of rules - generated by them.

121/n

The algorithm of evolution is straightforward and it produces things of enormous complexity.

122/n

Horse Gaits

Associated is this idea of - things like highly coordinated states of the body are like pattern forming processes that ripple through the entire body & often even are going through the environment with feedback & we have things like we've seen before like phase transitions.
123/n

As a horse speeds up over time, say it changes its gaits in very discrete ways - that's like a phase transition.

As it continues speed up it's velocity increases gradually but the gaits are changing abruptly.

124/n

We can think of complex coordinated behaviours as having these for one being generated as patterns that are globally generated through all the interactions across the system.

125/n

One reason that it's difficult/impossible to articulate what we do is because it's distributed throughout our whole body physically.

126/n

Where's the vantage point from which to articulate what's being done if all these different systems that are distributed across the body are participating in the generation of the pattern.

That's essentially why we can't articulate what we do.

127/n

You can walk but you can't say how you walk.

128/n

There's a sort of subfield of study called 'coordination dynamics' and they do some experiments to induce phase transitions in simple behaviours.

129/n

How do you steer a bike or a motorcycle?

Turns out most people can ride a bike but most people have no idea how they do it.

Why does the handlebar on the bike twist left and right?

130/n

Countersteering

Brian knows - push the handle bar that you want to turn toward.

You don't know it till you're steering something heavy, start riding a motorcycle then you learn, on a bicycle it's such a small sensation.

131/n

Another thing that comes up in motorcycle riding is target fixation

People get scared of hitting something
so they look at it
and so they hit it

132/n

You go where you look

133/n

It's interesting that the body is able to basically self-organise around your gaze.

What you're looking - at you're not telling your body what to do directly - if you were you wouldn't run into the thing

134/n

What you're gazing at is a constraint you're applying on the system and your body is self-organising to get you to that thing.

It's very, very different than most people assume it is.

135/n

One of the things you learn learning to ride a motorcycle is get your eyes off of that thing you're afraid of.

A lot of self-organisation, decentralised control, not how you think.

Look where you want to go.

136/n

Don't think of a pink elephant.

137/n

Also don't overdrive yourself, slow the f*** down if you're not comfortable.

138/n

In flying the danger is in the thing that doesn't move across your point of vision because either it's coming straight at you or straight away from you.

The angle is maintained when you're on a collision course which makes static it very hard to see in the first place.

139/n

There's problems with how scanning is taught in flying cause they tell you to look real hard for it but maybe what you should be doing is changing your speed or trajectory every now and again so that you can induce/notice something that's moving relative to you

140/n

The Air force is great at this

Civilians not so much.

141/n

There's a shocking amount of air to air collisions.
You'd think - 3D space, not that many planes - how would that happen?
It's exactly when things are on a collision course that they become harder to see.
The adaption of the retina can even work against you in some cases.
140/n

Other visual challenges:
Your retina is dynamic when something is static on it actually actively damps it out.
You literally could be blind to it & if something moves behind something that's static on your retina then it quickly damps it out - becomes invisible instantly.

141/n

142/n

One of the things that will keep coming up is this bandwidth constraint/limitations issue.
Each agent has limited bandwidth each system has limited bandwidth there's a constraint on just how much information you can process.

143/n

Additionally Patrick brings up: here's an example of Lung scans put across to lung cancer specialists who were asked to commentate look for outliers and just commentate on what they saw.

144/n

They'll commentate on the cancer nodules but 83% miss the dancing black gorilla.

145/n

Then when you actually look at scanning of where the actual eyes look on the image - a lot of them ignored it, but even some of them even still looked at the gorilla but still didn't actually realise that it was there.

A good example.

146/n

One of the dangers is around the expert - who is sort of pattern trained to look for things - is that they miss the outlier.

147/n

Another similar one - pilots in a simulator landing a plane - basically there's a huge jumbo jet just sitting right on the runway where they need to land.
Their attention is directed somehow but effectively they're looking at the runway see plane & they land right into it.
148/n

"Perception is a kind of skill and it's inseparable from other bodily skills".

149/n

Bringing a skill set to perception like the gorilla detection skill, (that they certainly have somewhere) it's just gone out the window while they're involved in this ongoing dynamically process of scanning for cancer.

150/n

Joe - 'I didn't see the gorilla.

Even though we were talking about it/primed - it took me a couple minutes to see.'

151/thread