Fired Up Neurons

1
Fired Up Neurons!

• Saturday Morning Physics
• December 18, 2004
• Presenter Rhonda Dzakpasu

2
What we know

• Simple elements of brain function
• Structure of brain
• Functional role of different brain structures
• Cellular composition of brain
• Action of neurons
• Action of neurotransmitters

3
What we dont know The Big Picture

• How does the brain WORK?!
• How does activity of neurons code behavior,
  cognition, memory?

4
Multiple Level Problem

• Bioinformatics what genes are involved to
  express proteins used in different aspects of
  cognition?
• Molecular approach
• Systems approach

5
Multiple Level Problem

• Bioinformatics approach
• Molecular what chemicals (e.g., ions,
  neurotransmitters) are involved in
• pathway needed for different aspects of
  cognition?
• Systems approach

6
Multiple Level Problem

• Bioinformatics approach
• Molecular approach
• System neuronal communication
• How do action potentials relate to cognition?

7
Is the Forest or the Trees?

• Static arrangement
• Everything is hardwired
• Stimulation of particular tree
• Thought corresponds to a particular tree
• Dynamical arrangement
• Ephemeral trees!
• Leaves form one arrangement and then change

8
Shes Baaaack!

• Static arrangement
• Young woman OR
• Old woman
• Not both!!

W.E. Hill
9
Two Faces or a Vase?
10
Many Sites are Activated

• Distributed information
• processing
• How different parts
• talk to each other

Courtesy of C. Ferris, K.Lahti, D. Olson, J.
King, Dept. of Psychiatry, Univ. Massachusetts,
Worcester, Mass.
11
Static or Dynamic?

• Static
• Need HUGE (infinite) forest for all
• thoughts!
• Dynamic
• How are the leaves functionally
• connected

12
Dynamic Communications

• How do the leaves on the trees
• communicate?
• An analogy Musicians in orchestra
• Practice is noise no communication
• When baton drops music to the ears!
• What is the difference between practice
• and play?
• Play correct notes at the same time
• - Notes, musicians are synchronized

13
But how does the brain work without a conductor?
14
Experimental ApproachOptical Imaging

• Optical imaging techniques convert
• information into light intensity fluctuations
• Monitor different regions of brain at the same
  time
• Study spatio-temporal structure of the dynamics
  of neuronal networks in vitro and in vivo
• fMRI not fast enough to detect action potentials

15
Optical Imaging

• Different types of signals can be imaged
• Intrinsic
• Chemical not used thats why intrinsic
• Low signal to noise must signal average
• Long time scale
• Dye-based Fluorescence
• Calcium concentration sensitive dyes
• Voltage sensitive dyes

16
Overview of Fluorescence
17
Fluorescence Excitation and Emission
Demo Time!
18
Fluorescence Imaging

• Voltage sensitive dyes
• Converts membrane potential into changes in
  fluorescence intensity
• Fast response
• Non specific

19
Fluorescence Imaging voltage sensitive dyes
20
Fluorescence Imaging voltage sensitive dyes
Ross, W.N., B.M. Salzberg, L.B. Cohen, A.
Grinvald, H.V. Davila, A.S. Waggoner, and C.H.
Wang (1977).
21
Fluorescence Imaging voltage sensitive dyes
22
Odor evoked oscillations in turtle olfactory bulb

• Objective how spatiotemporal patterns are
  changed when different stimuli is presented to
  sensory modality such as olfactory system

23
Olfactory System
nose
receptor cells
glomeruli
periglomerular cells
olfactory bulb
mitrial/tufted cells
granule cells
MTexcitatory GP inhibitory
24
Odor evoked oscillations in turtle olfactory bulb
Rostral
Caudal
Middle
25
Different cycles of oscillation employ different
neurons
1
3
2
10 isoamyl acetate
1
2
3
1 frame/4 ms
26
Period Doubling of Caudal Oscillation
27
Modeling the olfactory bulbWhat do we know?

• Three oscillations with different
• properties after the odorant
• presentation

28
Modeling the olfactory bulbWhat dont we know?

• Why do they form?
• What is their role in information
• processing?

29
Modeling the olfactory bulb
receptor cells
glomeruli
periglomerular cells
mitrial/tufted cells
granule cells
30
The Math behind the Model
Excitatory neurons
Inhibitory neurons
where
.
and
31
Modeling Odor Presentation
Interactions between cortex and olfactory bulb
32
Hypothesis Stemming from Model

• Two types of interactions are formed as a result
  of interactions between excitatory and inhibitory
  neurons
• They are phase shifted from what is observed
  experimentally

33
Hypothesis Stemming from Model

• Oscillations generated by excitatory neurons
  initially combine characteristics of the odorant
  expressed with the same strength
• Period doubling transitions observed only in
  caudal oscillation is reproduced by the model
  when the feedback from higher cortical regions is
  added

34
Modeling the olfactory bulb

• Simple anatomical assumptions of
• bulb
• Imitates behavior of bulb
• Imitates what the olfactory system does!

35
Turtle Signals

• Population recordings
• Thousands of neurons
• Signals are synchronized
• Like an orchestra playing a symphony

36
Single Neuronal Behavior

• What about individual neurons?
• What do individual instruments do when orchestra
  is synchronized

37
Temporal Neuronal Interactions and Memory

• Memory is formed by changes in
• synaptic activity
• Changes in synaptic activity depend on relative
  timing of action potentials

38
Temporal InteractionsNeurophysiology

• Long Term Potentiation and Long Term Depression
  as well as short term synaptic changes depend on
  the relative spike timings of the presynaptic and
  post-synaptic neurons

L.F. Abbott, S.B. Nelson (2000) Nature Neurosci.
39
Temporal InteractionsNeurophysiology

• In other words, synchrony and/or coherence
  between neurons underlies memory formation
• Here synchrony means the locking of action
  potentials

L.F. Abbott, S.B. Nelson (2000) Nature Neurosci.
40
Can we use analytical methods to measure how
neurons synchronize?
41
What is Synchronization?

• Adjustment of rhythms of
• oscillating objects due to their
• weak interactions.

SynchronizationA Universal Concept in nonlinear
sciences, Pikovsky, et. al., 2001
42
What is Synchronization in the Brain?

• Firing of action potentials at the same time or
  with preset phase
• Spatio-temporal patterns form
• Occurs in both healthy and non-healthy brain

43
Types of Synchronization

• Three types
• Complete or identical perfect linking of
  trajectories of coupled system
• Generalized Connecting output of one system to
  given function of output of
• other system

44
Types of Synchronization

• Phase perfect locking of phases of coupled
  system but amplitudes remain uncorrelated
• Occurs in non-identical and weakly coupled
  oscillator systems

45
Why Phase Synchronizationin the Brain?

• Neurons are weakly coupled
• non-identical oscillators

46
How do we measure phase synchronization?

• Identify a feature of a signal to study
• that can represent the specific
• value of the phase of the system
• Look for relationships between
• feature of interest that can define phase

47
How do we measure phase synchronization?

• Our feature time of
• action potential or
• spike
• Develop a measure
• based on changing
• list of relative spike
• times

48
How do we measure phase synchronization?

• Use this list to generate a distribution
• of probabilities of relative spike times
• Use entropy to evaluate properties of
• the probability distribution

49
What is Entropy?

• A system can be ordered or disordered
• Measure of randomness or uncertainty
• of a system

50
What is Entropy?
S - S p lnp
51
Lets Return to Neurons

• Since relative spike times are used, we
• say conditional entropies

52
Model Systems We Use
Rössler oscillators
Lorenz oscillators
Feature Poincare section z1
Feature Poincare section y0
Thalamocortical neurons (Hindmarsh-Rose)
Feature spike generation
53
Conditional EntropiesProperties

• Two coupled non-identical oscillators can phase
  synchronize
• The phase lag will depend on the relative
  properties of those oscillators namely
• If one unit has a higher frequency than the
  other, the other one will follow it and be phase
  locked

Black line neuron 1 Gray line neuron 2
54
Conditional EntropiesProperties

• The frequency mismatch in those oscillators will
  depend on their parameters
• Our measure will detect the direction of the
  phase lag between the two oscillators so that we
  can say which is following which

Black line neuron 1 Gray line neuron 2
55
Conditional EntropiesProperties

• Amplitudes uncorrelated
• (large synchronization error, exponentially
  decaying autocorrelation function)
• Phases correlated
• (large difference in CE between units)

56
Conditional EntropiesProperties
Real-time measurements of neural interactions
57
Conditional EntropiesProperties
In presence of noise
58
Conditional EntropiesProperties
Coupling strength
59
Synchronized but How?
Memory formation may occur when phase lag is
constant
60
Conditional Entropies and Memory
CEs can measure memory formation?
61
Monitoring SynchronyApplication to Epilepsy

• Changed structure of network to
• mimic axonal sprouting
• Spurious formation of
• excitatory synapses in injured
• area of the brain

62
Monitoring SynchronyApplication to Epilepsy

• Initially network is locally connected
• Randomly changed local connections
• to random global connections

63
Monitoring SynchronyApplication to Epilepsy

• We dont increase the number of
• connections just changed the
• connectivity of the network
• p 0 only local connections
• p 1 only random global
• connections in network

64
Monitoring SynchronyApplication to Epilepsy

• Based on conditional entropies
• we see how randomness in structure
• increases the degree of global
• synchronization in the network
• Global synchronization epileptic
• seizure

65
Monitoring SynchronyApplication to Epilepsy

• Phase synchrony as
• function of distance
• in the networks
• Varied the rewiring
• probabilities

Average distance between neurons (A.U.)
66
Monitoring SynchronyApplication to Epilepsy

• Local synchrony for low ps falls off with
  distance
• Global synchrony for high ps Stronger and
  distance independent

Average distance between neurons (A.U.)
67
Conclusions

• Systems approach to understanding behavior of the
  brain
• Use optical imaging with voltage sensitive dyes
  to monitor population behavior
• Use theoretical measures to predict and detect
  behavior of individual neurons within a network

68
Acknowledgements

• Zochowski Laboratory
• Michal Zochowski, PI
• Benjamin Singer
• Bethany Percha
• Soyoun Kim

Jonathan Edwards, MD Professor Department of
Neurology University of Michigan Hospital
69
Acknowledgements

• Timothy Chupp, Professor
• Jens Zorn, Professor
• Department of Physics

Lois Tiffany
Demonstration Lab Team Warren Smith Mark
Kennedy Harminder Sandhu
70
Acknowledgements

• My family
• Jasper, Noble and Philomena

71
Acknowledgements

• YOU !