| EEG_CONTEXT | Returns (in output 'delays') a matrix giving, for each event of specified ("target") type(s), the latency (in ms) to the Nth preceding and/or following urevents (if any) of specified ("neighbor") type(s). Return the target event and urevent numbers, the neighbor urevent numbers, and the values of specified urevent field(s) for each of the neighbor urevents. Uses the EEG.urevent structure, plus EEG.event().urevent pointers to it. If epoched data, also uses the EEG.epoch structure. For use in event-handling scripts and functions. |
| Usage: | >> [targs,urnbrs,urnbrtypes,delays,tfields,urnfields] = eeg_context(EEG,{targets},{neighbors},[positions],{fields},alltargs); |
| Required input: | |
EEG |
EEGLAB dataset structure containing EEG.event and EEG.urevent sub-structures |
| Optional inputs: | |
targets |
string or cell array of strings naming event type(s) of the specified target events {default | []: all events} |
neighbors |
string or cell array of strings naming event type(s) of the specified neighboring urevents {default | []: any neighboring events}. |
[positions] |
int vector giving the relative positions of 'neighbor' type urevents to return. Ex: [-3 -2 -1 0 1 2 3] -> return the previous 3, current, and succeeding 3 urevents of the specified {neighbor} types. [positions] values are arranged in ascending order before processing. {default | []: 1 = first succeeding} |
fields |
string or cell array of strings naming one or more (ur)event field(s) to return values for neighbor urevents. {default: no field info returned} |
alltargs |
string ('all'|[]) if 'all', return information about all target urevents, even those on which no epoch in the current dataset is centered. {default: [] -> only return information on epoch-centered target events} |
| Outputs: | |
targs |
size(ntargets,4) matrix giving the indices of target events in the event structure in column 1 and in the urevent structure in column 2. Column 3 gives the epoch number in which the target has latency 0 (else NaN if no such epoch). The fourth column gives the index of the target type in the {targets} cell array. |
urnbrs |
matrix of indices of "neighbor" events in the URevent structure (NaN if none). |
urnbrtypes |
int array giving the urnbrs event type indices in the {neighbor} cell array, else NaN if no such neighbor. Ex: If nbr = {'square','rt'} (see below), then urnbrtypes outputs are [1|2]; if nbr = {'rt'}, returns [1]s. |
delays |
matrix giving, for each {targets} type event, the latency of the delay (in ms) from each target event to its neighbor urevents. Else, returns NaN when no neighbor event. Output matrix size: (ntargets,length(positions)). |
tfields |
real or cell array of values of the requested (ur)event field(s) for the target events. Values are the same type as the field values, else NaN if no such event. |
urnfields |
real or cell array of values of the requested (ur)event field(s) for the neighbor urevents. Values are the same type as the field values, else NaN if no such event. If > 1 field specified, a 3-D array or cell array (nevents,nnbrpos,nfields). |
| Example: | >> target = 'square'; % target events are type 'square' >> nbr = {'square','rt'}; % neighbor events are either 'square' or 'rt' >> [trgs,urnbrs,urnbrtypes,delays,tflds,urnflds] = eeg_context(EEG,target,nbr,[-4 1],'position'); % % Output 'delays' now contains latencies (in ms) from each 'square' target event to the % 4th preceding and 1st succeeding 'rt' OR 'square' urevent (else NaN when none such). % Outputs 'tfields' and 'urnflds' give the 'position' field values of target events and % neighbor urevents. Output 'urnbrtypes', the index of the type (1='square' or 2='rt') % of the ('urnbrs') neighbor urevents. % >> trts = find(trgs(:,4)==2); % targets followed by an 'rt' (before any 'square') >> pos3 = find(trgfld = 3); % targets with 'position'=3 (numeric field value). >> selevents = intersect(trts,pos3); % target events by both criteria >> selepochs = trgs(selevents,3); % epoch numbers centered on the selected target events |
| Author: | Scott Makeig, SCCN, Institute for Neural Computation, UCSD, March 27, 2004 |