I have checked if there is any stimulus specific adaptation (SSA) in our sample, i.e. DOES THE RESPONSE OF A GIVEN NEURON FOR A GIVEN SHAPE DECREASES WHEN WE PRESENT THE SAME STIMULUS FOR 2ND, 3RD ETC TIME TO THE NEURON?

In order to see this, i compared the activity of my neurons for the 1st, 2nd etc. last (usually 8th or 10th) trials
for the shape evoking the largest response (BEST) in the basic condition (in case of my first study for COLORED shapes, in case of our second study, recov/noonrecov for the LINEDRAWING conditions).

I did it for the absolute (NOT THE NETTO!!!) firing rates after the start of the stimulation in a 500 msec window as well as for the ABSOLUTE NORMALIZED responses. (normalization is necessary for changes in basal firing rate across neurons DOES decrease  the result, to get rid of this- i normalized the data for each neuron separate and then i made an ANOVA across neurons (as Prof. Benedek has suggested).
Dependent variables were the ABS or NORMABS firing rates obtained during the 1st, 2nd etc repetitions, within subject factor was the number of repetitions (8 in case of REC/NREC, 10 in case of COL/LD study).

Here are the results:

There was no significant effect of repetition number IN EITHER STUDY for the ABSOLUTE firing rates:

COL-LD study, ABS firing rates. *******

oordinate:ABS firing rate in Hz, abscissa: repetition order, 1st, 2nd, etc repetition
effect: n.s.  (it is different from the first analisis results, due to the inclusion of more cells.)
figure

REC/NREC study ABS firing rates

oordinate abscissa as above
figure

now the same analisis for the normalized data: ********
it was ALMOST (!!!!!!) significant for the COL/LD study, meaning that here i could find difference in firing rate according to repetition rank order. (this is also changed since the last analisis. more cell are included. ergo the result is not so robust...)
COL/LD: ******
figure

however it was not significant for the other study:
 (n IS 21, and not 23-sorry)
REC/NREC:
figure

*******WRONG!!!
As it can be seen there is only significant effect of repetition in the COL study for the nromalized responses.
WRONG!!!
Even the normalised COL is only ALMOST significant...every effect is just at the border of the 0.05, meaning not too robust effects at all...

Now we went further, checked the same stuff for the more rarely seen shapes, i.e. for the COL/LD study for the LD ones.

absolute values: almost reaching level of significance at p<0.068
figure

normalised: this was significant at 0.02 ******
figure

The restrictions on the evaluation of the analisis: consider that although the NEURONS saw the stimuli DURING THE RECORDING OF THE CELL as it is stated on the figures, 1st, 2nd etc times, the ANIMAL HAVE SEEN IT ALREADY A THOUSAND TIMES DURING PREVIOUS DAYS OF RECORDINGS AND TRAININGS.   This means, that the NEURONS also must seen it a thousand times already, just not during the recording sessions.
ERGO: the sample is not 100 % suitable for this kind of analisis!!!

How could i avoid this potential problem? The COL in the COL/LD study, just like the LD in the REC/NREC study has indeed been seen by the animal a thousand times, but the LD and SIL in the COL/LD study, as well as the REC and NREC images of the REC/NREC study has only been seen by the animal a couple of times a day (when we run the test, intermixing COL/LD/SIL or LD/REC/NREC) this happened only 2-3-4 times a day (depending how many cells we could catch on the day).
For this reason i can make the same analisis as above for the LD and SIL, and for the REC, NREC images too.