Visualization

Visualization and Plotting

NeuralEmbedding provides ready-to-use plotting helpers for inspecting embeddings and neural activity. All plotting methods are instance methods on the NeuralEmbedding object and assume that embeddings (obj.E) and masks (aMask, cMask) are already configured.

Available plots

3D trajectory plot

• Call: NE.plot3(maxT); (default maxT = 40)
• What it shows: The first three embedding dimensions as colored 3D trajectories. A subset of trials closest to the median trajectory is plotted for clarity.
• Events: Add behavioral events before plotting with NE.addEvents(evtsStruct); (fields: Ts, name, trial, optional data for metadata). Events appear as scatter markers with a legend.

3D animation

• Call: NE.animate3(maxT); (default maxT = 40)
• What it shows: Animated 3D trajectories with a sliding trail (up to 50 time points). Useful for visualizing temporal evolution frame by frame.

Peri-event time histograms (PETH)

• Call: NE.peth();
• What it shows: Heatmaps of smoothed activity per area and condition. Requires homogeneous trial lengths; uses the original binned/smoothed data rather than the embedding.

Typical visualization workflow

% After embedding
NE.findEmbedding("PCA");

% Limit to a subset of conditions or areas if desired
NE.cMask = "ConditionA";
NE.aMask = "Motor";

% Plot trajectories
NE.plot3(25);           % static 3D plot
NE.animate3(25);        % animation

% Add events and replot
evts = struct( ...
    "Ts",    {0, 150}, ...           % one timestamp per struct element
    "name",  {"GoCue", "GoCue"}, ... % event labels
    "trial", {1, 2}, ...             % matching trial indices
    "data",  {[], []});              % optional metadata payloads
NE.addEvents(evts);
NE.plot3();             % events now overlaid

Tips

• plot3 and animate3 use the first three embedding dimensions; ensure the object NumPC property (or equivalent method parameter) is at least 3.
• Use masks to focus on specific conditions or areas before plotting to avoid clutter.
• peth is a quick diagnostic for preprocessing quality; run it prior to embedding to inspect firing patterns.