Cognition lectures 1-9

for a system to be computable and effective

finite set of components. (symbols) manipulated with a finite set of exact instructions or rules =. finite number of steps. ex of compiutable systems) chess. finite set of pieces and locations, rules for movement and capture. complicated but deterministic sequences of outcomes.

state spaces

offer a powerful tool for visualizing systems.

resting potential

-70 mv. inactive neuron are electrically excitable

how is neural computation accomplished?

1) synaptic connections modify the signal they transmit by a weight 2) neurons sum the weighted input of all of its incoming synaptic connections, add all up, sum of weight input 3) neurons compare their state to a threshold and iff the current state is greater than that threshold, create an action potential. determines whether a neuron fires or not

2 parts of hypothesis

1. mental processes are fundamentally symbol manipulation (character representation) 2. the symbols involved are implemented physically. hypothesis to get mind, representations have to be a physical symbol ex? writing notes

what do we need to represent

1. space. representations are thus mental states that mirror world states. surfaces and spaces. predators. awareness. 2. time: every mental activity is a rep. storage of and manipulation. rep. are unique to each person. present: immediate reactions. little way of rep. form basis for complex rep. past: learning and memory. future: complexities of language. predictions and anticipation

PSSH

2 in calculator. symbols and their manipulation must be the basis of any and all intelligent actions

what are representations

3 components. something (object reprepesented) 2. sign (representation itself). 3. somebody (someone the representation is meaningful to interpretant

how do we implement formal systems physically

computers. complex codes. how do we make computable systems physical? someone who does math. codes=representaiton. alan turing developed the basis for the modern computer

minds are computational and representational entities. what is computation?

Math. manipulation of numbers according to the rules of math. if rep. can be numeric, computations are formally constrained manipulation of rep.

multiple realizability

a given function, mental or behavioral, can be produced by multiple physical substrates (stuff, support, foundation)

spike train

a numeric signal (on or off, yes or no) that a given neuron sends to the neurons that it is connected to.

physical symbol hypothesis

a physical symbol system has the necessary and sufficient means for general intelligent action. representations are physical

state space diagrams by dynamical systems

a plot showing all possible states for a given system. banana dynamical system? every possible value of firmness. state space diagrams have one dimensoion for every dynamic variable used to define the system.

manipulations

anything that the mind does is accomplished by representations and/or manipulation of mind and its processes are mental representations. any thought is a representation

implicit

additional relationship. require other rep. or specific systems to use

refractory period

after an action potential neurons need to recover (equilibrium). neurons have 10-100 AP per second.

why do we need to represent the world?

ambigious input. the contents of mind mind consist of a copy of things in the world. imdivid objects, properties, concepts, desires.

what type of rep. is the armillary sphere?

analog and explicit

motor programs

analong: imagining. digital" actually doing something. human musculoskeletal system. redundant degrees of freedom. to move in a coordinated way we must organize and control these dgrees of freedom in a purpose. implicit representations require a specific system to function (ourselves)

What is the mind?

and organizational entity

godel numerbing

assign each symbol and function a unique natural number. ex) the way a computer represents characters (ASCII) more complicated version in order to ensure every symbol and operation was unique and represented. any formal system can be represented numerically and functions. Godel numbering shows that any formal system can be represented as a common set of symbols. computable=math

how does info get from world to body to mind?

avoiding action at a distance

synaptic connections are modifiable (they can change over time)

both weights and threshold can change based on experience w world. LTP: long term potential, increased responsiveness LTD: long term depression, decreased responsiveness. weights go down.

what defines a cognitive system

brain, body, environment, social environment

how we perceive, think, recall in similar ways

brains are all organized the same way

representations

can carry info from the world to the mind and allow us to interact with the world indirectly

symbols and signs

carriers of meaning to someone. can be literal symbols or meaningful relationships (fire)

receptor activations

cause changes in postsynaptic neurons that increase, EPSP, or decrease, IPSP, membrane potential of postsynaptic neuron. AP require activity in multiple synapses. chem released are how neurons communicate

human cognition/behavior

chruchlands crab. two eyes, two joints, limbs. eye positions and arm positions. this is a basic trigonometric relationship (more complex set of functions) sensory input transformed into motor output by computational function.

not all behavior is expressed by a simple linking input and output systems

complex problems can be made solvable if they can be broken down into a hierarchy of simpler problems. ex) smart behavior and replicating smart behavior with a collection of dumber behaviors. something start and breaking it down where it can be accomplished by units of yes or no. ex) reaching and picking up. complexity must emerge from simplicity. a system can be reduced is computable. if mind is computable and effective, we can avoid infinite regress.

what do we need to create a mechanistic account of reason? descartes thoughts

computable works with any symbols we want. 1) a formal system for arbitrary symbol manipulation. 2) a way to implement that system physically. 3) a way to represent mental states in that system

threshold

crossing threshold causes a spike

turing machine has a very limited set of capabilities

demonstrate how we can reduce more complex functions to sets of simpler ones. count. turing machine shows that complex functions, mathematical, or otherwise can be implemented in a simple mechanical way. shows that any numeric system can be implemented mechanically. combined with model numbering. mechanically is physical. turing machines determine if a given process or function is computable

dynamical systems and transistions

deterministic doesn't mean simple. complex behaviors even from low dimensional systems. slight change in initial conditions make it different. small changes in complexity.

godel

developed a way to represent formal systems mathematically.

memrane potential

difference in electric charge across the cell membrane. some move in and some out

Functions of the brain

different types of brains can produce behaviors with the same function (perception, memory, communication, coordination, language)

dynamical systems

dynamics are a way of describing the causal organization of a system. need two things. 1. distinct states. ex) being in lanigan, banana freshness, where they grow, bruised. turing machine would be values on tape and value and position of read head. 2. deterministic transitions between states. initial state system will move in exact same order each time but ending up in same ending state. system changes is determine by its state. known, fixed, decided

how do neurons communicate with each other?

electrically charged. sodium outside and positively charged. chloride inside and negatively charged.

what are relationships?

equivalent states

in different minds...

equivalent states correspond to each other because of how brain is related to the world, the rest of the brain, and the body

ambiguity

every aspect of perception. the info we receive at our sensory receptors is impoverished. things in the world are right side up, whole, and 3D. info at receptors is: upside down, broken into small pieces. 2D

emanation hypothesis

eye produces beams. beams contact objects in the world and return with a copy. when eyes open were perceive things. when not they can't

neurons

function is whether the neuron fires or not. basic representational and computational unit in the brain. timing and arrangement of a series of action potentials (spike train) are essentially numeric.

what matters the most

how components are organized

what the brain is doing

if the mind isn't the brain itself

why is turing machine impt to cognition

if we can express cog behaviors as the operations of simple components then we have hope of explaining all cog behaviors using a basic computational framework applicable to many systems. common framework between turing machine and brains

church turing thesis

if you can implement a function as a set of mechanical processes that system is effective and computable. if u can ,make a turing machine to do a given function that function is algorithmically computable. function can be replicated with a finite number of symbols. manipulated by a finite set of exact instructions that require no intelligence.

equivalent states?

interaction is primary focus, brain, body, and world

action at a distance

interaction of 2 objects separated by space without an intervening medium or mechanism (viewing things far away) series of events caused something to happen

how should we represent these things

interpretant symbol object. relationship should be causal. system activation of a particular rep. should be a reliable indicator of the object it signifies. rep. can be characterized on many dimensions

oppositie if deterministic

is stochastic. flipping a coin is stochastic. tules determine how turing machine changes states.

what makes a representation good

it must represent something in a faithful accurate way. relationship to event it represents must be consistent

what do we mean by somebody?

language of thought. what representations are and how they are manipulated and related to the world.

larger systems vs smaller systems

larger are more complex, have more components and are more difficult to approach theoretically. smaller may not be able to explain all of the phenomena of interest in a lawful way

a physical mechanism

like the brain. organized so that its states correspond to coordinates of objects and its transitions track the movement of those objects in the sky.

analog rep

literal copies. relationship is as faithful as possible. picturing an elephant

axon (output)

longer extensions of cell body connected to dendrites or cell bodies of other neurons. transmits signal via electrical and or chemical changes.

cognitive schience

mental and physical

representations

mental states that mirror world states symbols that allow us to interact with the world indirectly. everything in the head is supported by representation (hand drawing itself) allow us to make predictions about future and remember the past for both world and our system. carry meaning either internally or from the world. representations can be numerical (computation) the formally constrained manipulation of representations

physical system hypothesis

mind is a formal system, one approach is to demonstrate that its functions can be replicated by a purely mechanical system. mind can be reduced to a very simple computation. exact instructions=rules

representations of the world are not perfect copies

mind starts with ambiguous, inadequate representations. elaborate into faithfull representsentations. competition is one potential way this is accomplished

the mind is a system of functions that can produce goal directed behaviors

network of relationships

abstract relationships

not based on direct observations and guide behavior.humans are capable of determining. determine individuals and their past performance

human behavior is what

not deterministic but stochastic

digital rep

numeric. completely arbitrary. this is a continuum that captures how faithful the relationship between symbol and object is. word apple

eidola hypothesis

onjects in the world produce copies (intromission) we perceive water bottle because world shoots out copies

representations

perceiving an object we create a copy in our head. never the things themselves. anything mind is involved with represents and manipulates

can we approach the brain as a computational formal system?

physical system, like a turing machine. the brain is complex and hierarchically organized. start at high levels then go to smaller levels. neurons and synapses are relevant to computational understanding. we are focusing on computational and representational approach of the brain

dendrites (input )

postsynaptic. short extensions of cell body contains speaciliazed chem receptors. funnel signals

synaptic transmission

presynaptic terminal=axon terminal button. neurotransmitters are chem that connect with and activate specific receptors in postsynaptic neurons

mental properties

priveleged and physical. mind can be fundamentally material but physical prop are not reducible to same physical prop that define non mental matter

Descartes

rationalist. argued for the primacy of reason. reason being defined as the manipulation of meaningful symbols of rational rules. "ghost in the machine" all mental processes are the result of the minds rational operations on the states of the body. mind is the formal manipulation of physical symbols.

explicit

refer to object directly. the actual object

similar relationships?

relatinships are structured in a similar way

bonding relationships

relationships between things in system and the system. A and B have a bonding relationship if A changes as a result of its relation to B

computation of neurons

representation of neurons in the brain. asks a question when did the spike occur

multiple realizability

representations can have the same meaning in diverse physical forms. symbols can be abstract: they can take any form as long as that form constantly refers to a given object or event. this cabaility is what allows systems to represent infinity of possible states of world in an organized useful way

manipulating internal models

representations with identical causal structure to the world. (smoke to fire to tree). transitions from one internal state to another mirror the transitions between states of the external world. computation allows us to manipulate these models in order to plan and make predictions

mind body dualism

separation between physical and mental. mind and body are decomposable (broken down) can understand mind without worrying about states of the body. mind is special and distinct from physical.

trajectories

show how a system evolves over time (transitions)

what is mind?

similar brains doing similar things. physical property of the brain. Mind is the software of the brain.

degrees of freedom

this describes a physical system but if you use the wrong ones you'll be sorry. angle of wrist when reaching to grab something. you can do it in atoms but it would take forever

projective assumption

simulations are sent to the mind to be interpreted. implies that someone or something is doing the interpretations. (internal agent). brain creates simulations sent to mind and something does the hard work in head. this is how we avoid action at a distance. anticipation of behaviors.

medium hypothesis

something carries the copies to the organism. stuff out there we can't see but carries copies to us. minds are representational entities

simulative assumption

starting point for perception is a simulation (copy) of objects in the world. not actually seeing the world but seeing copies of it. copies mediate our interaction with the world.

neurons are the basic representational and computational unit

state (membrane potential) of a neuron is determined by its incoming connections (spiking or not spiking) how a neuron responds to the set of incoming spike trains determines its state and shapes its output to other neurons

what defines a system and a non system

systems are defined by bonding relationships

Mind is a system of functions that can produce goal directed behaviors

the mind is an organizational entity. we need a framework for systems, functions, and how they're organized regardless of how simple or complex they are. we need a framework that can handle groups of fruit as well as the complexities of the human mind

systems are defined by what

their boundaries

requirements to succeed

there are symbols. that can represent every possible or event in a meaningful way.

turing machine

tool for understanding computability. a general model for describing a computable and effective mapping between input and output. infinite one dimensional tape of data cells with finite states. 1 and 2. decrement: removing one. increment: adding one. this has a set of rules that specify what the components should do with a particular configuration. if machine is in current state and the cell being scanned contains symbol then move into next state and do action. yeses and nos. each machine has own sets of rules

armillary sphere

transform time info into place info. models the position of celestial objects, like the moon, over time to predict future locations. set the date using the outer ring, shows location of planets and objects on that date. creates a mapping between time and coordinate. transforms one into another. correspondence between two quantities, a function

Minds are organizational invariants

two minds implemented in different ways will be identical as long as they are organized the same. functionality is an organizational invariant

mind is a computational entity

under the idea of organizational. color=interaction of light source and properties of that. ex) lightness constancy. emitted intensity and surface reflectance. observer (I). I is the product of E and R. I can be produced by an infinite number of E and R pairs. I is ambiguous

hierarchal abstraction

we create scientific knowledge in levels. low and high levels. choose the right level of abstraction. in a hierarchy, each layer is defined by its relative complexity with more complex layers composed of multiple simpler layers. simplest level is army of idiots. human behavior=most complex.

boundaries

what is not in the system, components are ether inside or outside the system

action potential

when membrane potential reaches a certain threshold many channels open in rapid succession causing membrane potential to spike. ex) doors in mall on black friday. behavior of neuron is all or nothing. spike or no spike is yes or no. 0 or 1. synaptic transmission