Dayan: Theoretical Neuroscience

Following the book and the coursera course

28 flashcards · Biology · by lakesare

1. What are the potentials inside and outside the neuron during rest?

   Inside: -70mV Outside, in the surrounding bath: 0mV

2. What are 3 stages of polarization in a neuron?

   hyperpolarized (more negative polarization) polarized (exactly -70mV) depolarized (more positive polarization)

3. In which stage of polarization does the neuron generate an action potential? - hyperpolarized (more negative polarization)- polarized (exactly -70mV)- depolarized (more positive polarization)

   During depolarization, when voltage inside the neuron is a lot more positive than -70mV. If a neuron is depolarized sufficiently to raise the membrane potential above a threshold level, a positive feedback process is initiated, and the neuron generates an action potential.

4. Absolute VS relative refractory period?

   Absolute refractory period - when it's virtually impossible to generate a new action potential. A few ms.Relative refractory period - harder to generate a new action potential. 10s of ms.

5. What's a bilayer?

   A film 2 molecules thick, where each molecule is arranged with its hydrophobic end to the outside. Formed e.g. by lipids, in which each molecule is arranged with its hydrophobic end directed inwards towards the opposite side of the film and its hydrophilic end directed outwards.

6. Name 2 cases when you may want to have an electrical synapse.

   When we want 2 neurons to fire simultaneously When we want to 'synchronise' the fiting of two neuronsWhen we want an escape reflex E.g. found in a crayfish

7. What does a vertical bar mean on a raster plot? Vertical axis is each new experiment.

   A consistently strong response across trials (each new dot is a new trial). A vertical bar in the raster plot indicates a strong response, meaning that one cell fires very reliably at a particular time during the stimulus presentation, and suggesting that the cell is responding to the particular feature of the stimulus that occurs at that time. Weaker responses are represented by thin red bars (where the cell fires during some iterations but not others). As such, a weak response indicates that…

8. Encoding VS decoding?

9. What's the dimensionality of the vector?

   Number of elements in the vector.

10. What does the dot product do geometrically? (shadow-wise)

    Dot product is the project of x2 on x1, using the sun perpendicular to x1. THE BIGGER THE DOT PRODUCT, THE MORE ALIGNED THE VECTORS.

11. How would you think of this function as of a vector?

    If we subdivide the x-axis by, e.g., 0.1: f(x) = [0.05, 0.1, 0.15, 0.2, 0.25, ...] In the real world, the values that can take on are infinitely close together, but intuitively it’s easier to think of them as having some small spacing.

12. There is a sum operation on the vector. What would be a sum operation on a function? Sum(vector) Since functions are basically really high dimensional vectors, we can define vector-like operations on them.

    An integral of f(x).

13. There is a dot product operation on two vectors. What would be f(x) · g(x)? vector_1 · vector_2 Since functions are basically really high dimensional vectors, we can define vector-like operations on them.

    An integral of f(x) * g(x). This is a part of general intruition about functions - SUMS become integrals! Two functions are orthogonal if their inner product is zero.

14. What does the dot product of 2 functions tell you?

    How much 2 functions overlap.

15. When is f(x) · g(x) = 0?

    When the functions don't overlap.

16. [Linear filter] How could we extend this simplest coding model? As soon as there's a stimulus, there's a response.r(t) - responses(t) - stimulusϕ - constant

    We expect that r(t) depends not only on s(t), but on some combination of recent inputs: s(t), s(t-1), s(t-1), etc. The output signal of this system is linear in the input signal. This means that multiplying the input signal by a constant causes the output signal to be multiplied by the same constant, and that summing together several inputs yields an output equal to the sum of the outputs to each of the inputs individually. The fact that the system looks back several time steps does not prevent…

17. [Linear filter] How do we go from: to:

    τ acts as a dummy variable k.

18. When it comes to representing functions as vectors, sums always become...

    Integrals.

19. What does the LTI stand for?

    Linear Time-Invariant system

20. [Linear Time-Invariant System] What's input and output of the LTI?

    Input: a function Output: a function

21. [Linear Time-Invariant System] What's a good example of the LTI x(t) and y(t)?

    x(t) - a person speaking y(t) - a microphone recording Then:- when you speak a times as loud, the recording is a times as loud- when two people speak, it's equal to two recordings added up- this holds up through time

22. [Linear Time-Invariant System] What's h(t) called?

    Impulse response In the Coursera course, they call the impulse response function a filter.The impulse response is an especially important property of any LTI system. We can use it to describe an LTI system and predict its output for any input. Impulse response helps us find y(t) from x(t)!In signal processing, the impulse response, or impulse response function (IRF), of a dynamic system is its output when presented with a brief input signal, called an impulse.

23. [Linear Time-Invariant System] What's this operation called?

    Convolution. The output y(t) is 'the input x(t) convolved with h(t)'. Output is a 'filtered' version of the intput.Output is a 'convolution' of filter and input.To get the output, you convolve the filter with the input.

24. [Linear Time-Invariant System] How can you look at a filter f(t), and predict what y(t) will look like? Impulse response h(t), aka filter f(t)

    y(t) will be the highest when f(t) looks like x(t). Fun fact: filter is looking for the parts of the signal that look more or less like its own graph!Because of this, because our filter is looking for some features, we call our LTI a 'feature detector'.

25. What does the RC stand for in RC circuits?

    A resistor–capacitor circuit.

26. [Neural coding] Name 2 of the hypothesized coding schemes. Neural coding is a neuroscience field concerned with connecting the stimulus and the neuronal response.

    Rate coding - as intensity of the stimulus increases, the frequency of spikes increases. Under a rate coding assumption, any information possibly encoded in the temporal structure of the spike train is ignored.Rate coding is sometimes called frequency coding.Temporal coding - when information is communicated via the time between spikes. Binary symbols can be used to mark the spikes: 1 for a spike, 0 for no spike. Temporal coding allows the sequence 000111000111 to mean something different from…

27. [Neural coding] Rate coding VS temporal coding? Neural coding is a neuroscience field concerned with connecting the stimulus and the neuronal response.

    Rate coding - as intensity of the stimulus increases, the frequency of spikes increases. Under a rate coding assumption, any information possibly encoded in the temporal structure of the spike train is ignored.Rate coding is sometimes called frequency coding.Temporal coding - when information is communicated via the time between spikes. Binary symbols can be used to mark the spikes: 1 for a spike, 0 for no spike. Temporal coding allows the sequence 000111000111 to mean something different from…

28. r(t) expresses a linear combination s(t-2)f(2) + s(t-1)f(1) + etc. (s is a stimulus, and f is a constant)What else does r(t) remind you of? r(t) - responses(t) - stimulus

    The convolution operator.