Intro to Neuroinformatics

Aus Psycho
Wechseln zu: Navigation, Suche

140917 Intro to Neuroinformatics

Course Website: http://www.ini.uzh.ch/~fschuler/teaching/

Also list of books.
Principles of neroscience

Good book for models of descriptions theoretical "neuronomodels" very mathematical/physics ("Neurodynamics")

Exam Session 19.1-13.2 One exam for uzh and eth, it will be a written exam probably in science city

Understanding computation in the brain in order to improve computational systems. Study of the Brain with the help of computers

Brain

ca. 1.5 kg, 1.1-1.2 l Volume

Why do we have it?

D. Wolpert: The largest living thing doesn't have a brain (Giant sequia)

The Brain is for conscoious movement Influencing the World and interacting with it, only other way without moving to influence: Sweating

Sea squirt has nervous system can ove, but cant digest/feed, moves aroud, finds a rock lives on it and makes his brain into a digestive system

Transforming information from outside world into cognitive conscius movement.

Stimuli enters Brain, for this it has to be encoded , it transforms from the brain and encodes it in neural activity, the neural representation gives ability to perrception integration and moves into memory When we want to use this infomration we need to decode it into information/muscle movement.

What makes our Brain special?

We have a quite big brain (not the biggest though) the Shape also differs between species What makes the difference is a big debate but depends on the organisation of the Brain (elephant 3times Neurons, more in cerebellum, less in Cortex) Also relationship between brain and body size is different (humans are pretty high on this list)

How does Brainsize determine intelligence?

Neanderthal had bigger brain than we have, so we can say that the organisation makes a big difference. Neanderthal brain was probably very similar to sapines brain)

What can we do with our brains?

Construct things our brain hasn't changed that much over the time we can go into space with the same organ the neanderthals used! We got smarter by building better tools and making better use of those. Computers are a big thing here.

One thing we couldn't build without a computer is a comuter

Alan Turing

Building machines that are able to build better machines Turing machine (Theoretical construct) a machine processes information the automaton/machine (reads information and changes internal stge and is also able to write things to this tape. Universal turing machine: the code that is written is the code to build another turing machine Now one is able to develop a new machine that serves another purpose without changing the hardware, the only thing you change is the code.

John von Neumann

vonNeumann Architecture has one cantral Component (CPU) made out of two regions, control unit and Arithmetic logic unit CU keeps track where you are in Program, ALU does calculations, you also have external components, an important one is the memory (program an data are in memory!) basically the same as touring machine, but more practical and used in single core computers

Nowadays the bottleneck is not the CPU speed, but the speed of the memory access (von Neumann Bottleneck)

Neumann wrote "The Computer & the Brain" the frst perosn to make a comparison between Computer and brain. realized brain does not work like a compter

What's similar between a brain an a computer

  • process information
  • use logical oerpations
  • have memory
  • use electrica (digital) signaling
  • can learn from inputs
  • consumes energy

Where are the main differnces

if we where able to understand our brain we would be so simple thath we couldn't undertsand our brain

-massive parallelism constantly adapting chmical signalling unreliable units analog computation robust to damage very energy efficient

  1. Behaviur
    1. systems and pathways
    2. centers and local circuits
    3. neurons
    4. Microcircuits
    5. synapse
    6. membranes, molecules, ions

Number of operations compared to operating speed (old chart)

NI140917.1.png

Apollo control used really slow computers, enough to land on the moon

 Our Brain uses 20W of energy
 83000 Processors are able to blabla, Folie 25

Numbers of Nerve cells per organism on Slide26

Game to map neurons: eyewire

Human Brain Project

Goal: Simulationg the Human Brain on a supercomputer, hosted ad EPFL https://www.humanbrainproject.eu

Neurogrid

Investigate alternative computing structures that are emulationg the brain (aso opposed to simulating it like the HBP) Neuromorphic engineering You exploit what the physical system does by itself, save a lot of engery... Neurogrid chip of Stanford If you would simulate it you would need 100'000 times more energy.

IBM truenorth chip is also strongly influenced by neromorphic enineering, getting rid of the von neumann bottleneck.

SPAUN (Eliasmith)

More psychology based. Different approach, wiew the brain as interconnected boxes, which are different areas of processing, not trying to create a simulation, but design a system on the high level scheme that the brain uses and transform it into a network with nodes.

Cortex3D

Program that has implicit constructkion processes, you have simple primitives that do different things

http://www.ini.uzh.ch/~amw/seco/cx3d/

Chemical gradients. HTese Programs can be used to find out how the cortex develops (and how the lamination of the cortex takes place)

in lamination every cell runs sam program but the chamica conctntrations are different end let the cells specialize

Extremely complex simulation which runs on a very simple program, but it interacts alo t ith the environement

Neuroanatomy

Nervous System

CNS (Central Nervous system) & PNS (Peripheral NS), pretty complicated System, Anteric Nervous System (System of the Bowels) Good Tree, Subdivision on Slide2 Planes of Brain bisections (Horizontal, Coronal, Sagital, Good Graphics Slide33 Directions of Orientation ond Slide4 Bone Structures of the Skull Slide5 Brain is in the meninges and surrounded with CSF Development of Nervous syste, Slide7 Brain flots in Liuquids and Ventricles are also filled (CSF is produced in Ventricle) Nice Graphics of Major division around10

Limbic System

Broca's libic lobe (was the fist to describe it), Connection of Structures. Associated with emotion, Memory aquistion. (in the 50's amygdalas where removed)

Hippocampus

Circuitry of hiocampus pretty laid out, a lot of wht it does are part of how the cells are connected with one another, it's there to store information into long term memory (Patient HM, epilepsy, remove Hippocampus) He could not learn anything new for the rest of his life. Don't remove the Hippocampus on both sides. Motor learning etc. still worked, just the semantic memory was busted

Hypothalamus and Thalamus

Thalamus contains a lot of the sensory relays nuclei, Domain specific Processing,

Hypothalamus

Collection of Nuclei that look thath the Body functions correctly, controlls autonomic machanisms, Vascualr disease, Wernicke-Korsakoff syndrome, Thalamic Trauma

Basal Gangia

Series of Nuclei, which are mostly for voluntary movement, subcortical nuclei Smooth well organized sequences of movment, we don't know everything they do, are related to Parkinsons disease. Substancie Nigra, produziert Dopamin, stark bei Parkinsons!

Brain is extremely interlinked, all the Systems wor closely together.

Structures in moor control: many different areas are involcves (cerebral Cortex Some of the Fingermovements have one sinapse in spinal cord and goes to Finger directly from Cortex, Basal Ganglia and Cerebelum work together with the cerebral Cortex.

Cerebellum

involved in Movement,, layered does postural adjustmens and is very important for motor learning. And a Flow of Motor movements

Reticuar Formation

Diffusa Arrangement od Nuclei and Neurons Arousal and selective attention, it Regulates Wake and Sleep cycle. People in a coma often tried to get peaple out and has sometimes wrked is to electrically stimulate the formatio reticualris Anesthetics often work here, uncounsciusness

Cerebral Cortex

Lobes

In the middle you hacve white matter and nucleii WM are just axons, surfce is the Cortex with its lobes, gyri and fissures. The cerebral cortex changes in it's organisatiom

Hte Main Components we are interested in are the Neurons

  • Cell body
  • Nucleuas
  • Dendrite
    • Antennas (receive Signals) Are not only passive cunductors, but can also ba active and conduct AP's
  • Axon
    • many neurons have axons , in the Retina the Neurons dont have axons, not all Neurons have axons (long cable)
  • Myelin
    • Isolation
  • Boutions
    • Contacts are made on Boutons (Terminal Boutons)
    • Endknöpfchen

Pain is transferred in 2 Waves you also have C-Fiber whcih transfer Pain much slower. Is the Pain that comes later. Axon is specialized name for a particular fiber in the neuron, output fiber input fiber is dendrite.

A cell has one Axon, whcih can Brach out. Very intricate Arbrisation, you can have even more comlex Dendrite Arborisation

Anterograde vs. retrograde axonal transpot

Signal transmission is mostly elecrical, if it gets to the synapse, release of a neurotransmitor Some neurons are directly connected electrically, gap-junctions (Photoreceptors are often connected in gap junctios)

We estimate 1000 Billion Neurons in the Human Brain, it's an extremely complex organ

Synapses

Made out of boutons, cleft, dendritic spines etc. Synapses dont always have to be on the dendrites but there are many possible wasys to connect it.

It's possible to change efficacy of contacts between strenghts and efficiency (weight of a synapse)  you can change what the network does. A lot our brain can so is because synspases can change their efficacy, but you can create/remove synapses within seconds. Brain is constantly changing how it's wired. Central mechanism how we learn
synaptic plasticity many different ways on how this could work, many possible machanisms.
Most effects on our Brain (throug Drugs) happen at the Synapse

At the moment we pretty much understand the biological and electrical workings of the cell.

Different animals have very different neurologies, the neurology of a neuron can have an impact on it's function The morphology of the Neuron has an effect on it's funtion

The electrical Properties are very different.

Nissl stains

the pattern of connection is very important . Cortex is organized in different layers (nissl stain means that you stain the tissue for different cell parts) Peaople are trying to find out why we hve different layers in our cortex

Golgi methods

another method to Stain cell Bodies

Cytoarchitecture

looking at organisation of the Cortex, 6 Layers of the Cortex Dendrites make connections at particular levels, you can try to find ourt what computations are performed based on the connections.

Brodman Areas

What areas are connected to waht others. Labeled 50 Areas of cortex based on anatomical differences (which als mean different functions) Abnormal connections (synestesia) (Vladimir nabukov) big part of it is genetics. One of the theories are abnormal connections between areas, we don't understan dho it works exactly

CA1 Region of Hippocampus

Hippocampus produces rythmic Signals, can be produced by this circuit. If you record someones EEG brain is rythmically firing. where does this come from? Probably hippocampus CA1

Next circuit slideis from NI institute All cortical areas can be broken down in many modules, which all do the same basic computation, based on the same circuit. big debate Is the same pattern of connection available in all cortexes.

Reciprocal Innervatiomn of antagonistic muscles, Intricate Circuitry with feedback signals Circuitry that automatically tells the muscles to work this way! Motor behaviour is the most important thing to communicate.