hi, I have a very basic question. I record with tetrodes in freely moving rats to study how the neurons respond to the cue, I got the firing rates for one single unit and plot the peristimulus histogram of the firing rates. Now I want to see whether the neurons respond to the cue, do you know what statistics I should use and how long the time window before and after the cue I should select? I checked some papers,but it seems the methods are variable. THanks for your ideas,
An option is to compute a z-score, namely subtracting the baseline mean from your PSTH, and then dividing it by the baseline standard deviation. If your z-scored PSTH goes above or below 2, it means your cell's firing rate went 2 standard deviation away from the mean. Thus, your cell can be considered significantly excited or inhibited. The chosen time window depends on the kinetic of your responses.
Hi Marco,
Thanks for your answer. I already calculated z-score. I have a 100 ms bins for plotting PSTH. The question is how many bins with a z-score of more than 2 there should be for me to judge whether it is significantly excited or inhibited.
I think this depends on the physiology of the neurons you record from, the responses you are getting and it is largely subjective, as in the case of the choice of the bin size.
Dear Tiayang, I think that 100 ms is a very large time window to calculate a PSTH of firing neurons. In this work that I share you, I did it with 1 ms bins! However, as Marco says, it is a very subjective criteria and you should fix that bin-window according to the features of your data....
Article Fast estimation of motion from selected populations of retin...
Thanks Alexander! It is true 100 ms is a very large time window. Now I am trying with a criteria that at least 2 bins in 1 s of post-event with z-score more than 1.96 or less than -1.96 to detect significance. But I think the criteria is based on your data.
Tianyang, do keep in mind that you will run into the problem of using multiple comparisons:
Using a 5% significance level (roughly corresponds to your z-score threshold) by definition 1 in 20 bins will be false positives. So if you require one bin out of 10 (100 ms bin width, 1 second post event) in a single experiment to be significant, you can see that your risk of a false positive becomes very high. Setting the criterion to 2 bins out of 10 reduces this error, but you are still not anywhere close to a 5% significance value that you actually intend to use.
Ultimately, you will need to find a way of either correcting for the multiple comparisons, or testing your whole PSTH in a single comparison.
Also, using such wide bins, you risk not seeing brief increases. I had an experiment in the past where neurons showed a very strong but very brief (under 5 ms) time-locked response. When I initially used 100 ms bins, the response just averaged out. When I moved to 5 ms bins the peak was clearly visible and reproducible. As always, 'your mileage may vary' but it is something to keep in mind.
Good luck with your experiments!
Hi Christian, may I ask you in which brain region and in what situation you got this strong and time-locked response?
Sure. I was using light airpuffs to the upper lip region, under isoflurane anesthesia. Whole-cell and cell-attached recordings in cerebellar lobule crus-II.
No doubt very different from what you are doing. If you are expecting slow, delayed responses, longer bins will be fine. But it is worth testing. No matter what, it will be important to correct for multiple comparisons.
Good luck!
Hi Marco Bocchio,
how do you compute mean and standard deviation of your neuron firing rate? what kind of binning do you use?
thank you!
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis