- “various types of behaviour could not be explained in terms of stimulus-response mechanisms. Instead, psychologists need to think about organisms as storing and processing information”
- Assumptions of behaviourist:
- Learning is a result of conditioning
- Conditioning depends on association and reinforcement
- Tolman and Honzik
- Latent learning, rats without reward learning the maze
- What form does this latent learning take? is it spatial , are they remembering paths or some knowledge about how the maze fits together etc.
- Using place representation as opposed to reinforcement. Fig 1.3. One of the first proposals to represent behaviour in terms of representations
- Lashley
- We should think about many behaviours a products of planning and organization. There is a hierarchy to the planning. A large task is broken into smaller subtasks.
- All of this being done subconsciously
- Hypothesis of subconscious information processing. e.g. the calculations required to move your arm to pick up a bottle of water
- Hypothesis of tasks analysis understanding of a complex task by breaking it down. Cognitive systems can be understood as performing particular tasks. For example, the memory system allows the exploitation of recently acquired information. Its use can be broken down into sub task after sub task.
- Turing Universal computer, I still dont understand this
- Linguistics, Chomsky
- Why languages have the structure they do not just categorising language forms and syntax
- Analyzing sentences in terms of their phrase structure
- Sophisticated, hierarchically organized, cognitive ability such as speaking and understanding a language, involves stored bodies of information (about phrase structures an transformation rules)
- Bodies of information can be manipulated algorithmically
- Miller
- Basic concept of information theory, the information channel. A medium that transmits information from sender to receiver
- Miller used infromation channels to describe perceptual systems (vision etc.
- Seems to be an upper bound to the number of distinct items that can be processed simultaneously
- We have information bottleknecks
- Chunking helps us to optimise what we can process (binary vs decimal representation of a number)
- Natural language is the ultimate chunking tool
- representation as the basic currency for cognitive science
The implications of the Turing machine
- If we except that the mind is an amalgamation of information processing components then the Turing completeness means that any algorithm can be performed, so any information processing component can be simulated or replicated.
- Its important to define what a Turing machine is first to determine its capabilities
- Tape: Cells of symbols that the computer can read or write
- Head: Reads and writes symbols
- State Register: Replace the ‘state of mind’ according to Turing. Stores the state of the Turing machine
- Table: Tells the computer what to do for some current state and symbol observed
- Every part of the machine is finite apart from the tape
- Church Turing hypothesis
-
A function on the natural numbers can be calculated by an effective method if and only if it is computable by a Turing machine
- There is no process for computation we can perform outside of the Turing machine to make something calculable
- An identification of computability of a function with the effective calculability of that function
-
- Interesting example could be mapping this to problems or solutions we have to problems (ideas) in programming
Entscheidungsproblem
-
algorithm to decide whether a given statement is provable from the axioms using the rules of logic.
- Given the notion of effectively calculable as that described above A general solution to the Entscheidungsproblem is impossible
- No such program (one that can determine if an algorithm will halt (is decidable) or not can exist) because of the ability to create a paradox within our current system.