Phys735- lab exercise #3.

Auditory Nerve Fiber Single Unit Analysis.

Sept. 23, due Oct 1


Read the data analysis portion of Computers in Neurophysiology (P. 11 to end).


Each class member is to analyze an auditory nerve fiber's responses to sounds that cover the unit’s response area. 

See assignment table at end of document.  You will need the IDs listed there to retrieve the data.


You can use the PCs in room 87 for this exercise. Double click on the ST340 icon. Then login on the VAX (username =phys735, password = phys735).

(you can also log on the VAX in room 81 directly. Type mvl.  Username =  phys735. then password = phys735).  Type lo to logout.


The analysis package name is RAP which stands for response area program.

 type RAP (enter)


at the prompt: type DF phys735L3. specifies the file that contains the 'raw data'.  The data consists of the times of occurrence of discharges of auditory nerve fibers that were recorded from a chinchilla’s auditory nerve with a micropipette filled with 3molar KCl.  The only information about the axon’s behavior was the time at which the action potential (voltage) passed a criterion level set by the experimenter on an oscilloscope using the trigger level.  The times are recorded to 1 ms accuracy. These times represent a complete description of the ANFs response to the sounds that were presented.


Each unit has three separate items that are to be studied: a different unit is assigned to each student. Data to be analyzed is specified using the id command, e.g., id 1-1 selects the first response area data set (see table p3).  


1. response area: the response of a unit to repeated presentations of a tone over the frequency-intensity plane has been called a response area.  The 'area' is sampled for a specified range and in a specified increment for the two parameters, in this case frequency and level.  Note: the term response area for the output is actually a misnomer and since the output is really a response surface, sometimes the graph is called a response map.  This output will appear in the standard plot you will generate.

use a log axis for the abscissa. (hint type log xx)

Graph the spikes vs frequency.   type gr sp


From this plot you can determine the ANF's CF, threshold, spontaneous rate, maximum discharge rate and it's Q10. (Q10 = CF/(bandwidth at 10 dB above threshold). These numbers are to be recorded in the Class summary description of lab3.


Try graphing the response to only one repetition of the stimulus. (hint: type tr 1 1 to analyze only the first trial. tr 3 4 to analyze trials 3 to 4, etc.). Comments?

Smooth the curves: type sm line n.  This is a symmetric moving average filter (e.g., n=3 means y(i)=(x(i-1) + x(i) + x(i+l))/3:

1) why is it a symmetric filter and what does that do for you?

2)        What is the restriction on n?

3)          Could you use this filter to smooth a signal in real time?  If not, why not?



generate a 3D plot: type gr sp 3D


One of the commonly used analysis techniques is the IRC or isorate curve.  This has an equivalency with the frequency threshold curve (FTC) for a unit.  FTCs are often gathered using an adaptive up/down procedure to determine the threshold at each frequency.  We will use the response area data.


type gr IRC.


you can (first) specify the criteria for which the curve is determined (e.g. PER = 10). This is the percentage of the maximum rate minus the spon rate.

This analysis usually works better when the data is smoothed ahead of time.  Measure the

Slope (dBs/octave) of the IRC above and below CF (you need a hardcopy first & a log scale). Record these values.

 Given the maximum discharge rate for the unit.

4) what change in rate does the 10% criteria correspond to? 

5) How many spikes in a 50ms tone pip would this correspond to?


List the points that determine the curve: (type di  irc or pr irc) you can then enter this data into a MATLAB file or Excel spread sheet so that all the curves can be superimposed in a single plot (a population study).


Before leaving this topic consider the tradeoffs between using an FTC to obtain CF, TH, Q10 and an isorate curve obtained from a response map.

6) What are some of the tradeoffs?


2.           Short-tone at CF (use the proper data set) In order to study the statistics of the unit ( & obtain smooth appearing plots), the short tone is repeated 250Xs.  The unit response can be studied using Poststimulus time histograms (PSTH), interval histograms (IH), cycle histograms (CH), latency dot display (LDD) and a variety of other techniques.  Plot each of these and examine the outputs. Reset the number of bins used for analysis…. Type nb 500. what is the temporal resolution (us/bin)?


type:    gr pst, gr ch, gr isi and gr ldd.



7) when should you 'believe' the phase value obtained from the calculation of vector

strength that is provided in the cycle histogram?



3.           Rate -level function.  Another method of studying unit response is to determine how they respond as some parameter is varied- e.g., stimulus level which results in a rate-level function.  plot spike rate vs level, the synchronization vs level, and latency versus level.

type:  gr sp, gr lat, gr syn


execute the macro: em  phys735rlf

there are four curves generated…. Each has its own ordinate.


It would be interesting to combine all synchronization coefficients vs frequency

curves in a single plot.




Each student: hand in (Oct. 1)  RA plot, IRC curve with slope calculation, PSTH, IH, rate level curve.

 Answer all the questions in this exercise.  Repeat each question before you answer. Word processing documents only…. No hand written responses.


The class is responsible for:

The superposition of all IRCs in a single graph (abscissa: log10(frequency)).

Summary graphs of unit thresholds, Q10, spontaneous rate, maximum discharge rate, slopes of the IRC curve. Make sure the curves are labeled adequately.




Chinchilla auditory nerve fiber assignment.



(1) response area: IDs

(2) short tones:  IDs

(3) rate curve: IDs