CGSC170 Lecture Notes - Lecture 9: Karl Lashley, Activation Function, Biological Neural Network
Document Summary
What is learning: we need an operational definition (cid:862)a(cid:374) e(cid:448)e(cid:374)t p(cid:396)o(cid:448)okes a (cid:374)e(cid:396)(cid:448)ous s(cid:455)ste(cid:373) alte(cid:396)atio(cid:374) that p(cid:396)o(cid:448)okes a (cid:272)ha(cid:374)ge i(cid:374) (cid:271)eha(cid:448)io(cid:396)(cid:863) learning = a change in brain state triggered by some event. Induced lesions in rat/monkey brains: lesions affected the acquisition and retrieval of memories, but not storage. Memories are not stored in discrete locations: memories are distributed, stored across the network, a lesio(cid:374) i(cid:374) a si(cid:374)gle lo(cid:272)atio(cid:374) does(cid:374)"t e(cid:396)ase a si(cid:374)gle (cid:373)e(cid:373)o(cid:396)(cid:455) Neurons link together in network: linked neurons form neural circuit, activation of the circuit generates the experience of the memory. If two connected neurons are activate at the same time, their connection strengthens. Increase in connection = if one is activated, increased likelihood of other being activated. Synaptic plasticity: ability of synapses to change their structure or chemistry. Remember learning in anns: ann components, nodes, links, weights, thresholds, nodes = soma (neuron body, links = axons (neuron tail, threshold = activation function.