Week 5 - LTM

Different types of LTM

Declarative (knowing that) and procedural (knowing how). Within declarative there is episodic (memories) and semantic whereas procedural encompasses skills mostly

Complementary learning systems - Kumaran et al 2016

Damage to the hippocampus

Impairs learning of new associations (found in case of HM)

Causes problems retrieving earlier memories, particularly recent ones - Squire 1992

Preserves ability to recall more distant memories, performance on intelligence tests and normal conversation

Preserves ability to understand general meaning - Knowlton et al 1992

Semantic memory

General knowledge of the world, abstracted from individual experiences. Used to make inferences and predictions about the world. (for example hear something bark, assume it is a dog, might bite, might be chasing a cat, could be a cavachon)

Schemes and scripts in semantic memory

Types of knowledge structures used to organise and understand information - Bartlett

Symbolic approach to semantic memory

Knowledge represented in a mental language consisting of propositions. propositions made up of concepts and relations. Semantic networks represent knowledge spatially as units linked by relations

Semantic network - Collins and Quillian

Problems with symbolic approach

Typicality effects (Rosch 1973)- Some instances are more typical of a categpry than others for example a robin is more typically a bird than a penguin

Rosch 1976 - Some levels are more salient than others - Intermediate level categories are associated with more attributes - basic level categories. For example contrast furniture with a chair

More problems with symbolic approach

Concepts vary with context - what attributes does a frog have changes if they are in a French restaurant

Concepts are theory dependent - unclean animals vs clean animals

No mechanism for learning - how can these networks change as we learn

Parallel distributed processing - PDP

Information represented by ‘dumb’ neurons which just add activation and pass it on
Information is distributed as patterns of activation across many units
Learning is accomplished by changing the strength of connections
between units

Learning in PDP networks - backpropagation of error

Input: Patterns corresponding to various examples of cats and dogs
Required output: CAT or DOG
Each input pattern is presented
Activation is passed through the network
Output is compared to target (error)
Weights are adjusted to reduce error
Next example is processed, and so on until all
patterns are classified correctly

Distributed representations in the brain - Kiani et al 2007

we examine responses of a large number (>600) of neurons in monkey inferior temporal (IT) cortex with a large number (>1,000) of natural and artificial object images. During the recordings, the monkeys performed a passive fixation task. We found that the categorical structure of objects is represented by the pattern of activity distributed over the cell population. Animate and inanimate objects created distinguishable clusters in the population code. The global category of animate objects was divided into bodies, hands, and faces. Faces were divided into primate and nonprimate faces, and the primate-face group was divided into human and monkey faces. Bodies of human, birds, and four-limb animals clustered together, whereas lower animals such as fish, reptile, and insects made another cluster.

Semantic dementia

Gradual loss of semantic distinctions - reverse of early development, move from specific to general such as dog used for different animals