New User Special Price Expires in

Let's log you in.

Sign in with Facebook


Don't have a StudySoup account? Create one here!


Create a StudySoup account

Be part of our community, it's free to join!

Sign up with Facebook


Create your account
By creating an account you agree to StudySoup's terms and conditions and privacy policy

Already have a StudySoup account? Login here

CIS 140 Week 3

by: Alexis Mitchnick

CIS 140 Week 3 CIS 140

Alexis Mitchnick

Preview These Notes for FREE

Get a free preview of these Notes, just enter your email below.

Unlock Preview
Unlock Preview

Preview these materials now for free

Why put in your email? Get access to more of this material and other relevant free materials for your school

View Preview

About this Document

Covers perception and modularity, ANNs, visual pathways
Intro to Cognitive Science
David Hoyt Brainard, Lyle H Ungar
Class Notes
cognitive, Science
25 ?




Popular in Intro to Cognitive Science

Popular in Cognitive Science

This 6 page Class Notes was uploaded by Alexis Mitchnick on Friday September 16, 2016. The Class Notes belongs to CIS 140 at University of Pennsylvania taught by David Hoyt Brainard, Lyle H Ungar in Fall 2016. Since its upload, it has received 27 views. For similar materials see Intro to Cognitive Science in Cognitive Science at University of Pennsylvania.

Popular in Cognitive Science


Reviews for CIS 140 Week 3


Report this Material


What is Karma?


Karma is the currency of StudySoup.

You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!

Date Created: 09/16/16
Tuesday, September 13, 2016 CIS 140 Week 3 Connection and Modularity - Review of Marr Levels - Estimate probability using Bayes rule — Representational - Ripe fruit by color — Computational - Describe world using list of symbols — Representational - Use “tape” with “read head” to read symbols — Hardware (if describing physical tape), Representational (if describing abstract Turing Machine) - * ‘using’ or ‘by’ can be key words for representational - Brains and Computers - Brains: use distributed representations, many modules and sub-modules - Making computer like a brain: brains don't work like TM’s - use artificial neuron networks (connectionist models) - parallel analog computation, distributed representation - Architecture matters — unlike in TM, memory, computation speed, and reliability matter - Brains can still be modeled by a TM — TM can compute anything a brain can.. but TM seems more “fragile” than a brain does (an entire computer program can be ruined by the change of just one character) - Modeling a brain - Artificial Neuron Networks (ANNs)— connectionist models - Many interconnected “neurons” - Parallel Distribution Processing (PDP) - “neurons” learn from neighbors, take input from others and produce output - Computer vs. Brain — - centralized vs. distributive computing 1 Tuesday, September 13, 2016 - sequential vs. parallel processing - fast vs. slow - separation vs. integration of memory and computing - explicit programming vs. learning - Modern-day computers are looking more and more like brains, don't fit same model as described: - ex. Google — failure problem: want a piece to be able to fail without entire system failing - much cheaper to buy thousands of processors than one highly advanced one - heat and power is more efficient when distributed throughout — one central chip would get extremely hot - write only one algorithm that helps program to “learn” (ex. FB recognizing faces, didn't write program to recognize each individual face) - Computer circuits are faster:10 state changes per second (10 bytes is a Gigabyte, 6 10 bytes is a Megabyte) 3 - Human circuits are slower: (10 state changes per second) - Artificial Neural Networks (ANN): - highly simplified models of neurons - combined in layers - calculate features of inputs ‘in parallel’ - learn by adjusting weights - different types: Backpropgation, Radial Basis Function (RBF), Hopfield, Boltzmann, etc. - acts as an abstract mapping device - weights equivalent to connections - Nodes: neuron-like processing elements - Links: synapse-like connections between nodes - Biological Neuron: neuro transmitters — dopamine, serotonin 2 Tuesday, September 13, 2016 - Single Sigmoidal Artificial Neuron: - each node takes in the summation of all the surrounding outputs times the weight of each connection - “a million of these together” allows machine to start to act like it is smart - Input goes into nodes, then gets passed to hidden processing nodes, then produces output in last nodes - ANN does pattern recognition - ex. image of person —> who it is; picture —> caption describing it; a word —> the sound of saying it; sound of a word —> the word; sequence of images ‘seen’ by car —> how to turn - OCHRE ANN Demo: - neural network architecture - challenge of neural memorization - learning and generalization — memorizing things exactly can sometimes be useless, must generalized (because nothing you see is the same as the last time you saw it) - overfitting - Can map anything with ANN (“deep learning”): ex. can map a person walking to a cat walking — creates a person walking like a cat - deep learning applications: game playing, object recognition, machine translation, speech to text and vise versa - Example vision network: mapping —> image —> over 1,000 labels - What you perceive is a function of what you expect (learning is based on what you've seen): this is why computer program ‘thought’ a boom box was a cellphone - Brain Simulation: image —> auto encoder —> auto encoder —> auto encoder - As you go deeper into network: devotes more neurons to things that are frequent, network learns structure of images - Current limitations of ANNs: - don’t represent complex structures well (ex. logic issues) 3 Tuesday, September 13, 2016 - don’t scale as well as brains - Localist vs. Distributed representation of concepts - localist or “grandmother cell” — firing of individual neurons or small groups represent concepts (same neuron(s) fire every time you see your grandmother) - distributed or holistic — representation by pattern of firings or large sets of neurons - localist view does not scale, our brains are distributive - Distributive is better for mistakes than localist — gives some notion of how certain you are (car ex. — should you go left, right, or straight ahead? distributed pattern is more informative) - better generalization - represents certainty - improves accuracy - Red Bicycle Problem: - Extreme Localism: one neuron recognizes ‘red’, another ‘bicycle’, a third ‘red bicycle’ - Problem: either impossibly many very specialized neurons or must address binding problem - Concept representation: - compromise between specific and general - distributive representation - the binding problem - Universal constraints on computation: - can’t remember everything, pay attention to everything, can’t connect all neurons together Modularity cont’d - Visual Pathways Overview: 4 Tuesday, September 13, 2016 - optic nerve —> —> primary visual cortex (responsible for perception) - axons — control dilating and contracting of pupil to control light - signals we’ll focus on: back in visual cortex - Visual Fields and Cortical Hemispheres - if a person is staring straight ahead, can distinguish between a left visual field and a right visual field - In Left Eye: RVF —> Temporal Retina (light ends up on the side closer to temples, on the outside); LVF —> Nasal Retina (light ends up on the side closer to the nose, on the inside) - In Right Eye: RVF —> Nasal Retina; LVF —> Temporal Retina - *Both eyes receive information/light from both visual fields - For Both Eyes: Temporal Retina: stays on same side; Nasal Retina: crosses over - Net Effect: Both eyes see both visual fields, for either eye — LVF —> right side of brain; RVF —> left side of brain - Visual Field Test — test one eye at a time; spots flash at random locations, subject presses a button when he/she sees a spot, diagram result: black area represents area of visual field where subject did not see flashes (black indicates blindness — normal vision = no black) - Modularities in visual processing — info. from left and right visual fields is initially processed separately by left and right hemispheres (aka — understanding visual experience doesn't necessarily give insight into how we can experience a “seamless visual world”) - Anatomy of brain: white matter carries signals, grey matter does the processing; inward fold = sulcus, outward fold = gyrus - Brodman — examined different dye staining patterns in brain to create a cytotechtonic map of brain - MRI Scanner: used today to non-invasively examine brain, creates images - three views: saggital (profile view), coronal (back view), axial (overhead view) 5 Tuesday, September 13, 2016 - can create 3D virtual renditions of the structure of the brain based on MRI images - structural MRI measures for water density in brain - functional MRI (fMRI) assesses neural activity by measuring blood oxygen levels - BOLD Signal: shows modulation waves of vascular activity as it relates to spikes in specific neural activity - Hierarchical Elaboration of Response Selectivities Across Visual Areas - Ventral Pathway (“what”) - Dorsal Pathway (“where”) - Patients with Lesions — - Patient with Damage on Ventral Pathway: deficits in processing, problems judging orientation (couldn’t recognize orientation of line when trying to match index card with slot)… experiment was slightly altered and patient did OK (lesson: couldn’t perform a “what” task, but could perform a “where” task) - Patient with Damage on Dorsal Pathway: difficulty perceive motion (couldn’t pour tea/coffee or tell when to stop pouring) (couldn’t perform a “where” task, but could perform a “what” task) - Double Dissociation: provides evidence for separate functions being processed in two areas (as opposed to single dissociation, which is damage to one area that affects one function — harder to prove separate functions this way) - Different paths/areas in brain respond strongly to different things: - Fusiform Face Area (FFA): strong response to faces but not places - Parahippocampal Place Area (PPA): strong response to places but not faces - Another example of double dissociation — two different areas of brain have strong responses to one stimulus and not the other 6


Buy Material

Are you sure you want to buy this material for

25 Karma

Buy Material

BOOM! Enjoy Your Free Notes!

We've added these Notes to your profile, click here to view them now.


You're already Subscribed!

Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'

Why people love StudySoup

Jim McGreen Ohio University

"Knowing I can count on the Elite Notetaker in my class allows me to focus on what the professor is saying instead of just scribbling notes the whole time and falling behind."

Janice Dongeun University of Washington

"I used the money I made selling my notes & study guides to pay for spring break in Olympia, Washington...which was Sweet!"

Steve Martinelli UC Los Angeles

"There's no way I would have passed my Organic Chemistry class this semester without the notes and study guides I got from StudySoup."


"Their 'Elite Notetakers' are making over $1,200/month in sales by creating high quality content that helps their classmates in a time of need."

Become an Elite Notetaker and start selling your notes online!

Refund Policy


All subscriptions to StudySoup are paid in full at the time of subscribing. To change your credit card information or to cancel your subscription, go to "Edit Settings". All credit card information will be available there. If you should decide to cancel your subscription, it will continue to be valid until the next payment period, as all payments for the current period were made in advance. For special circumstances, please email


StudySoup has more than 1 million course-specific study resources to help students study smarter. If you’re having trouble finding what you’re looking for, our customer support team can help you find what you need! Feel free to contact them here:

Recurring Subscriptions: If you have canceled your recurring subscription on the day of renewal and have not downloaded any documents, you may request a refund by submitting an email to

Satisfaction Guarantee: If you’re not satisfied with your subscription, you can contact us for further help. Contact must be made within 3 business days of your subscription purchase and your refund request will be subject for review.

Please Note: Refunds can never be provided more than 30 days after the initial purchase date regardless of your activity on the site.