N6 - How Can Perception Be Affected - Either By Current State, Or By Beliefs

Front 1

Does cognition just happen in the brain or does it also take into account bodily cues i.e. embodiment of cognitions?

Is cognition functional or objective?

Back 1

Perception can be objective, but often can be more useful in the long run if it's not simply objective and equal

Perception of the environment cannot be separated from the perception of the body and people will perceive the world depending on how they're going ot act on it (Gibson 1979)

Perception is functional - found some cells in retina that fire for large, slow objects i.e. predators, and some that fire for small, fast objects i.e. prey... seems complex for the retina, but it's adaptive (Ingle 1973)

Front 2

Proffitt's experiments and theories?

Back 2

Hills appear steeper if you're wearing a backpack∴ perception taking into account demands (which are greater if there's a heavy load)

Hills also appear steeper if you're tired or elderly∴ perception taking into account capabilities too (lower if old/tired)

(Proffitt et al 1999)

Perception overall has to budget resources properly to ensure you don't spend more (i.e. demands of task) energy than you have (i.e. capabilities) = Economy of Action theory

Can take into account physical and social resources as well as demands of task ahead and use this to adjust perception of world so as to bias action path to maximise survival.

(Proffitt 2006)

Front 3

What are Schnall's experiments?

Back 3

Shows hill steepness only affected on visual/verbal measures, not haptic→ different streams? Dorsal stream for haptic, about action planning∴ needs to be acccurate, because it feeds into motor with parieto-motor circuits.

Showed lower steepness if:

• Sugar drink (as opposed to sweetner)
• Friend present - correlates with how long you've known the person
• Imaginary friend (as opposed to somebody who betrayed you or neutral)

(Schnall et al 2008, 2010)

Front 4

Can impending action/intention also affect perception?

Back 4

Examples of impending action/intention manipulations:

• Arm flexion (approach cue) or extension (avoidance cue)
• Maze task: Get cheese (approach) or run from predator (avoid) or neutral (nothing)
• Told they will have to climb hill (approach) or not (avoid)

If poor physical condition → less effect ∵ if its not possible, then approach cue contradicted by physical difficulty?

Also combining two e.g. maze, then told they'll have to climb → much bigger effect

(Krpan & Schnall 2014)

Front 5

What is Social Power and what signals it?

Back 5

Social power is the force in relationships that enables hierarchies - can be signalled by availability and control of resources

Front 6

What happens if you're powerless as opposed to powerful?

How can you measure power?

Back 6

Powerlessness can lead to less cognitive flexibility, but increased attention to tasks∴ perform better at discrim tasks e.g. vigilant processing task (Gerinate 2007, Weick et al 2011). Power can affect perception - increased attention.

Can be measured - personal sense of power test.

Front 7

How can you manipulate power and how does that affect perception?

Back 7

Examples of power manips:

• Experience High/Low Power Postures
  
• Imagine Low/Neutral/High situation
  

If asked to estimate weight→ power manip→ low power→ heavier (Lee and Schnall 2014)

Can even affect physiology e.g. cardiovascular response (Scheepus et al 2012)

Front 8

On neural architecture of perception and how that might be affected?

Back 8

Traditionally thought that perceptions are transduced into amodal representations e.g. feature list, but now seems more likely that it uses an analogue model → could be more easily affected by taskdemands/capabilitystate/intentions (as demoed above)

Could also just be feedback from higher levels though, because processing not just serial - also the breakdown of visual stimuli does happen at the information entry, but it's rebuilt into a more complete picture at higher levels of visual streams.

Front 9

Can prior beliefs (and other things) shape perception - how is this important in Psychosis?

Back 9

Psychosis = loss of contact with reality. Can either be a symptom of a disorder (e.g. schizophrenia), or part of a healthy patients life (e.g. hearing voices)

How can some people live in their own made up world of hallucinations (faulty percepts) and delusions (faulty beliefs) that are actually very complex and complete. For it to feel so real, must not be a pathology in interpretation of cues, but in the percepts themselves?

Percepts are actually generated from within i.e. not just objective recreation of reality, but subject to a lot of changes before it becomes conscious:

• Original sensory input
• Affected by prior beliefs - to try and make sense of imperfect inputs
  
• Affected by biases e.g. demands / capabilities / intentions - to ensure 'economy of action'
  

Psychosis as complex, complete, pervasive and effective as there are can only emerge if perception is generated by the brain. And something goes wrong in that generation process.

Front 10

Why does a healthy brain need to model the world?

Back 10

The point of the brain is homeostasis, and it can mitigate changes in the world better if it can predict those changes. It can predict changes better by modelling the world. (Conant & Ashby 1970)

Front 11

What's the best way to model the world?

Back 11

The best way to model the world is using evidence from the world! i.e. sensory inputs. Given sensory inputs, we can infer what the world is like (abductive inference)

But sensory inputs are noisy/ambiguous, so we add prior beliefs i.e. previous models of the world in there.

→ Bayesian inference of the world = What is the likeliest cause of these inputs given what I know before.

Danger in using prior models - kinda moves us away from reality, but actually if our model keeps updating, should be ok

Front 12

What is an important aspect of modelling reality with prior beliefs that means we don't lose contact with reality?

Back 12

Sensitive to a mismatch (prediction error) between prior belief and what actually happened in the world → allows us to update the model and build it correctly.

But isn't this a logical quandry, because the prediction error means you compare the sensory input and the model and force model to be like input? So then how can model affect inputs → perception?! Actually you only change model in certain circumstances (see next card), and generally the model and input work together→ perception

Front 13

What two factors make the prediction error more intelligent.

Back 13

The prediction error relies on comparing model with sensory input (which makes the logic a bit circular because the model also affects the output), but we also weigh up the reliability of the model with the reliability of the inputs. i.e. if input is more fuzzy - put more weight on model. If situation is new, then put more weight on inputs. Dynamically change what we rely on. If relying on model, don't update much based on unreliable input, if relying on input, do update because model is poor.

Can also use the magnitude of difference to assess if model ought to be updated i.e. if it's not that far from the model, then maybe update the model (e.g. grass got shorter - cut), but if it's crazy - seek more information without necessarily updating model (e.g. grass suddenly becomes purple). Magic is when inputs and expectations collide.

Front 14

Examples of prior shaping perception

Back 14

Binocular Rivalry = separate image presented to each eye - perception only picks one that dominates, although can switch. Interestingly works with half in each eye (tells you that selection happening at/after visual areas that are binocular). Works: Prior knows that it can't be half house, half face, so picks one or the other.

Gaze adaptation - adapt to face looking left → biases straight/slightly offset gaze to right. UNLESS told that the first face is blind∴ doesn't bias nearly as much. (Teufel et al 2009)∴ must be quite high level control of bias.... i.e. high level prior can shape unconscious adaptations too!

Random dot kinegrams - appears to move as a cyclinder, but switches direction at random (bistable perception). If glasses put on that bias direction (e.g. by red/green), then identical glasses given (without the physical bias)→ still biased in reporting that glasses direction. (Sterzer et al 2008). I.e. perception being shaped by prior of glasses.

Last two also examples of updating model i.e. updating model of person you see, updating model with glasses

Front 15

How does this work neurobiologically?

Back 15

High levels of brain hierarchy transmit downstream∴ percept goes through to upper level once modified AND prediction error signal generated here if model needs updating.

Front 16

So what does the model affecting perception / prediction error affecting model mean for psychosis?

Back 16

If there's a problem with the prediction error signal, then the model of the world could become increasingly abberant→ can lead to delusions/hallucinations because model is so strange that the priors bias the perception totally wrong→ new reality

As the model differs from the world too → more prediction error, but if the prediction error system is broken→ probably makes it worse (otherwise should be correcting)→ even more bizarre model = runaway.