Trial representations

This module defines the basic classes that represent trials in Elephant.

Many neuroscience methods rely on the concept of repeated trials to improve the estimate of quantities measured from the data. In the simplest case, results from multiple trials are averaged, in other scenarios more intricate steps must be taken in order to pool information from each repetition of a trial. Typically, trials are considered as fixed time intervals tied to a specific event in the experiment, such as the onset of a stimulus.

Neo does not impose a specific way in which trials are to be represented. A natural way to represent trials is to have a neo.Block containing multiple neo.Segment objects, each representing the data of one trial. Another popular option is to store trials as lists of lists, where the outer refers to individual lists, and inner lists contain Neo data objects (neo.SpikeTrain and neo.AnalogSignal containing individual data of each trial.

The classes of this module abstract from these individual data representations by introducing a set of Trials classes with a common API. These classes are initialized by a supported way of structuring trials, e.g., TrialsFromBlock for the first method described above. Internally, Trials class will not convert this representation, but provide access to data in specific trials (e.g., all spike trains in trial 5) or general information about the trial structure (e.g., how many trials are there?) via a fixed API. Trials are consecutively numbered, starting at a trial ID of 0.

In the release, the classes TrialsFromBlock and TrialsFromLists provide this unified way to access trial data.

TrialsFromBlock(block, **kwargs)	This class implements support for handling trials from neo.Block.
TrialsFromLists(list_of_trials, **kwargs)	This class implements support for handling trials from list of lists.