Announcements• a3 is out, due 2/13 and 2/22 11:59pm•Computational: part 1 isn't too hard; part 2 is harder• quiz will be graded in about a week.Where we stand• Last Week–Psych experiments–Imaging studies –Connectionist representation• This Week–Learning–Backprop• Coming up–Neurophysiology of colorThe Big (and complicated) PictureCognition and LanguageComputationS tructured ConnectionismComputational NeurobiologyBiologyMidtermQuizFinalsNeural DevelopmentTriangle NodesNeural Net & LearningSpatial RelationMotor ControlMetaphorSHRUTIGrammarabstractionRegier ModelBailey ModelNarayanan ModelChang ModelVisual SystemPsycholinguistics ExperimentsBrain study techniques• What is it, and how is it used?–lesion studies–PET–fMRI–single-unit recordings–EEG and MEG– EcoG– TMSSpatial and Temporal ResolutionRepresentations• What is a localist representation?• What is a distributed representation?• How many things can you represent with 4 neurons, in each representation?• How many conjunctions of things can each represent?• What is coarse coding?• What is coarse-fine coding?Coarse Codinginfo you can encode with one fine resolution unit = info you can with a few coarse resolution unitsNow as long as we need fewer coarse units total, we’re goodCoarse-Fine Codingbut we can run into ghost “images”Feature 2e.g. Direction of MotionFeature 1e.g. OrientationYXGGY-OrientationX-OrientationY-Dir X-DirCoarse in F2, Fine in F1Coarse in F1, Fine in F2Biological memory1. Into what categories do people divide memory?–Long term versus short term versus working–Skill versus declarative (semantic and episodic)Declarative Non-DeclarativeEpisodic Semantic ProceduralMemoryTwo ways of looking at memory:facts about a situationgeneral facts skillsMemoryShort Term Memory Long Term MemoryTwo ways of looking at memory:electrical changesstructural changesLTPBiological learning1. What is retrograde messaging?2. What is the biological mechanism for short-term memory? Long-term memory?• Hebb’s Rule: neurons that fire together wire together• Long Term Potentiation (LTP) is the biological basis of Hebb’s Rule• Calcium channels are the key mechanismLTP and Hebb’s RulestrengthenweakenWhy is Hebb’s rule incomplete?• here’s a contrived example:• should you “punish” all the connections?tastebud tastes rotten eats food gets sickdrinks waterWith high-frequency stimulation, Calcium comes inDuring normal low-frequency trans-mission, glutamate interacts with NMDA and non-NMDA (AMPA) and metabotropic receptors.Recruitment learning• What is recruitment learning?• Why do we need it in our story?• How does it relate to triangle nodes?Models of Learning• Hebbian ~ coincidence• Recruitment ~ one trial• Supervised ~ correction (backprop)• Reinforcement ~ delayed reward• Unsupervised ~
View Full Document