Band pass in delay estimator

[lepmik]

If I understand the delay estimator correctly it estimates the delay on the entire high frequency part of the PAC (dar model) (assuming the driver is low freq and the pac high frequencies e.g. theta gamma). Would it be possible/easy to implement a way to define which high frequency band to estimate delay for?

I'm asking because I typically get PAC looking like: (from entorhinal cortex L2)

When estimating delay, I'm not sure if it corresponds to the high frequency blob or the low frequency blob.

[TomDLT]

Interesting figure !

This is definitely a limitation of the toolbox, due to the fact that the design is centered around DAR models. In DAR models, the modulation of amplitude is estimated on all frequencies at the same time, so the delay is unique.

In your data, you seem to have different PAC behaviors in the band [10, 30] Hz and in the band [60, 100] Hz. If you want to estimate two delays for the two bands, you will need to implement your own AddDriverDelay, adding a filtering step to separate the two bands, and apply two different delays.

Importantly, you need to consider the possibility that this is not a delay, but a phase shift (see this example to distinguish the two notions).