Once the raw data have been transformed into P1 and P2 values at a specific inter-pulse delay, the data can be processed into more meaningful forms. These procedures essentially follow those outlined in previous studies [Joy et al, 1989; 1990].
After the first pulse reaches the granule cells, an inhibitory phase is triggered, which persists for some milliseconds, slowly decaying, which is seen as the difference in amplitude between the second pulse-evoked peak (P2) and the first (P1). The IC50 measurement is the inter-pulse delay over which the evoked inhibition has decreased such that P2 is 50% of P1; thus the I50 represents the time in milliseconds at which the inhibition of the second pulse has reached 50% of maximum. Using the center point of the decay of inhibition proves to be quite accurate, in a similar way to LD50 or EC50s.
One calculates IC50s by collecting series of paired-pulse potentials. The initial delay is small, but long enough that the pulses overlap, usually > 20msec. An intuitive guess is usually made at what will be a reasonable delay. With each new paired pulse the delay is increased in steps. A gap of around a second is left between each paired-pulse to allow the inhibitory phase to completely pass and the preparation to recover. Once a series of paired pulses have been applied, these can be analysed to find the peaks, and the P2 can be expressed as a percentage of P1 and plotted against the inter-pulse delay. 
It can be seen that inhibition is best at short inter-pulse delays; as the delay increases paired-pulse potentiation rapidly compensates and produces P2s greater than P1.
Once a series of I50s have been calculated, these can be plotted against the time at which their requisite data were measured. One then has an interpretation of how the I50 changes over time from when a pyrethroid in given to the animal, in turn reflecting how the recurrent inhibition is being affected by pyrethroids. Many sets of results like this, from different preparations, can be plotted alongside each other, revealing trends.
When comparing data from different preparations one should be aware of the propensity for individual variations between each animal. These variations can be moderated in some ways, for instance by removing the baseline from each data set. One can do this by simply subtracting the initial pre-dose readings from every I50 value. There are more sophisticated approaches, such as expressing each value as an increase relative to the baseline, thus taking into account the variation in relative signal strengths. However more sophisticated manipulations warrant preliminary work to verify their expediency.
The I50 data were accumulated at irregular intervals, due to the technical nature of the experiment. It is thus not possible to directly compare and co-analyse values from different groups. To get around this we binned the data into 5-minute segments, rounding to the nearest 5 minutes. This small bin size proved accurate, but left some empty bins.
For some analysis the empty bins were not problematic, however in generating smooth curves empty bins would have offset the curve. To deal with this a second data set was created, in which empty bins were given a value interpolated from the two nearest bins. This method is far from perfect, as it still normalises the data, but at least it does so to a lesser degree. With the sets of binned data mean plots could be made for each treatment group.
The dual-treatment group received esbiol at +45 minutes, so that the peak effect would coincide with that of deltamethrin, following the procedure used in vitro [Song et al, 1996]. it was necessary to offset the esbiol time bins to reflect this before adding them to the deltamethrin values. This new group of generated values for esbiol + deltamethrin could then be plotted against the real values.
[Joy et al, 1989; 1990], [Song et al, 1996]
