pseudopop¶

Build a pseudopopulation by combining units across sessions after condition-wise trial averaging.

What this operation means¶

pseudopop takes a list of session DataArray objects and, for each session:

Computes a PSTH grouped by one or more trial-condition coordinates (group_by).
Concatenates the resulting condition-averaged units across sessions along the unit dimension.
Adds a session coordinate on the unit dimension so each unit retains session provenance.

This is useful when no single recording contains enough units, and you want a single condition-aligned population representation for downstream analyses (for example PCA, dPCA, GPFA, or decoding on condition means).

Example¶

from aind_ephys_utils import pseudopop

pp = pseudopop(
    [session_a, session_b, session_c],
    group_by="choice",
)

# pp has units from all sessions and a condition dimension ("choice")
# plus a "session" coordinate attached to each unit.

Notes¶

Input arrays should share compatible condition definitions used in group_by.
Trial-level variability is intentionally removed by averaging within each condition before concatenation.
If session_ids is omitted, session labels default to s0, s1, …