2D decoder place_fields are a fixed value along the Y dimension

[blakeporterneuro]

Version 1.3.15

When using the 2D decoder, I was getting odd decodes.

I plotted the place fields decoder.plot_place_fields()
and saw my fields had stripes in the Y dimension. I checked and the values along the Y dimension are all identical.

[edeno]

You're giving it 2D position of shape (n_time, 2)?

[blakeporterneuro]

yeah, ndarray, n_time, 2

Array

Plotted

[blakeporterneuro]

Shedding a little more light, the stripes only occur if you use the 'spiking_likelihood_glm' algo and does not occur if you use the 'spiking_likelihood_kde'. Though your paper implies the GLM is what should be used for sorted spikes?

_kde example

[edeno]

Ah sorry I did mean to investigate this. The KDE is fine here. I think the issue could be knot placement for the splines in the 2D case, which is sometimes tricky.