Data analysis - Stellar Prospector

From TACWiki
Revision as of 09:34, 13 June 2011 by Belmonte (Talk | contribs) (New page: '''Stellar Prospector''' The game begins with SP_Phase1Begin (9) and comprises three phases, in this order:<br> Phase 1: Visual cue-change signal. Targets only, no distractors.<br> Phase...)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Stellar Prospector

The game begins with SP_Phase1Begin (9) and comprises three phases, in this order:
Phase 1: Visual cue-change signal. Targets only, no distractors.
Phase 2: Visual cue-change signal. Targets and distractors.
Phase 3: Auditory-visual cue-change signal. Targets and distractors.

The task is to respond with a down-arrow key-press to high-value targets at fixation, whilst simultaneously monitoring for lower-value targets in the four sectors surrounding fixation in the upper visual hemifield, and responding with an up-arrow key-press to these. At any given time, exactly one of the four sectors is the cued sector, in which 80% of targets will appear.

Depending on the numbers of trials available, data can be pooled across Phases 2 and 3, or even across all three phases. (However, this matter of data pooling is for the experimenters to decide; the processing software should keep the phase-marker events intact.) For translation of context-sensitive to context-free event codes:

Original Event following event classification
SP_CentralStimulusSpawn SP_CentralStimulusResponsePressed CentralStimulus=active hit
SP_PeripheralStimulusResponsePressed CentralStimulus=active response error
SP_MissedCentralStimulus miss
SP_PeripheralStimulusSpawn SP_CentralStimulusResponsePressed CentralStimulus=inactive response error
SP_PeripheralStimulusResponsePressed PeripheralStimulus=active hit
SP_MissedPeripheralStimulus miss
SP_DistractorSpawn SP_PeripheralStimulusResponsePressed PeripheralStimulus=inactivefalse alarm
SP_DistractorDisappear correct rejection

Also re-code the associated response into a hit, a response error, or a miss based on the re-coding of the stimulus. Response events that are not associated with any stimulus by the above rules and occur with PeripheralStimulus=inactive should be re-coded as false alarms. (These will be quite rare in Phase 1, but more common in Phases 2 and 3 which involve distractors.)

For each of the three phases separately, and also for phases 2 and 3 pooled without phase 1, compute the following:
Trial-by-trial values, mean, and standard deviation of:
Reaction times to central stimuli
Reaction times to validly cued peripheral stimuli (where the value of the Sector parameter for the SP_PeripheralStimulusSpawn event is the same as the value of the Sector parameter for the most recent SP_CueChange)
Reaction times to all invalidly cued peripheral stimuli

When listing out the trial-by-trial values for the peripheral stimuli, also list the stimulus's polar coordinates (r, θ), the number of the sector in which the stimulus appeared, and the number of the currently cued sector (from the most recent SP_CueChange). Use this information to compute multiple regression coefficients of reaction time against the stimulus's r (its distance from fixation) and against the absolute value of the difference between the stimulus's θ and the θ that runs along the centre of the cued sector (-3π/8, -π/8, pi;/8, 3pi;/8; in general the formula is Sector×π/4 - 3π/8). Also compute statistical significances for these regression coefficients.

For the peripheral stimluli only (not the central stimuli), in Phases 2 and 3 only (exlcuding Phase 1), in two separate computations one for the currently cued sector and the other for the currently uncued sectors, complute d' ("d prime") using the tallies of hits, correct rejections, false alarms and misses. (If any of these tallies is zero, estimate d' by increasing it to 1.)

1 SP_PeripheralStimulusResponsePressed
    PeripheralStimulus={inactive, active}
    CentralStimulus={inactive, active}
    Gather={inactive, active}
The up-arrow key has been pressed, indicating a response to a peripheral stimulus (as this response might be a false alarm, a peripheral stimulus may or may not actually have been active); the parameters indicate the current activity of the central and any peripheral stimuli, and whether the game will be responsive to this key press or whether it's temporarily inactive as a penalty for a previous, incorrect key press.
2 SP_CentralStimulusResponsePressed
    PeripheralStimulus={inactive, active}
    CentralStimulus={inactive, active}
    Gather={inactive, active}
The down-arrow key has been pressed, indicating a response to the central (fixation) stimulus (as this response might be a false alarm, the central stimulus may or may not actually have been active); the parameters indicate the current activity of the central and any peripheral stimuli, and whether the game will be responsive to this key press or whether it's temporarily inactive as a penalty for a previous, incorrect key press.
5, 6 SP_Start_Menu_Action At the beginning of the game, a selection has been made from the menu. (These codes should not be being transcribed onto the parallel port and thus should not be appearing in the EEG log.)
7 SP_MenuSelectButtonPressed The player has pressed the space bar in response to an instruction prompt
9 SP_Phase1Begin Phase 1 of the game has begun: visual cues only (no simultaneous auditory cue), targets only (no distractors)
10 SP_Phase2Begin Phase 2 of the game has begun: visual cues only (no simultaneous auditory cue), 50% targets and 50% distractors
11 SP_Phase3Begin Phase 3 of the game has begun: simultaneous visual and auditory cues, 50% targets and 50% distractors
12 SP_End The game has ended; the player has collected enough peripheral stimuli (and ejected enough central stimuli) to fill the power bar
13 SP_PeripheralStimulusSpawn Sector={0, 1, 2, 3}
    {X, Y}=screen coordinates
    ThetaRadians=angle from fixation in polar coordinates [—&pi/2;, &pi/2;]
    RadialDistanceInPixels=display-coordinates distance from fixation
    ScreenDistanceInCm=linear distance from fixation (cm)
    RhoRadians=anglular distance from fixation (radians)
    AzimuthPhiRadians=azimuth from fixation (radians)
    ElevationLambdaRadians=elevation from fixation (radians)
A lower-value stimulus in one of the four peripheral sectors, away from fixation, has appeared;
the parameters give distance from the fixation point in several redundant coordinate systems
14 SP_CentralStimulusSpawn The high-value stimulus at the fixation point has activated
15 SP_DistractorSpawn
    Sector={0, 1, 2, 3}
    {X, Y}=screen coordinates
    ThetaRadians=angle from fixation in polar coordinates [—&pi/2;, &pi/2;]
    RadialDistanceInPixels=display-coordinates distance from fixation
    ScreenDistanceInCm=linear distance from fixation (cm)
    RhoRadians=anglular distance from fixation (radians)
    AzimuthPhiRadians=azimuth from fixation (radians)
    ElevationLambdaRadians=elevation from fixation (radians)
A distractor in one of the four peripheral sectors, away from fixation, has appeared;
the parameters give distance from the fixation point in several redundant coordinate systems
16 SP_CueChange
    Sector={0, 1, 2, 3}
    Sound={true, false}
The visual cue has changed to the designated sector;
it is accompanied by an auditory cue iff Sound=true
17 SP_MissedPeripheralStimulus The currently active peripheral (low-value) stimulus has timed out and disappeared without a response
18 SP_MissedCentralStimulus The currently active central (high-value) stimulus has timed out and disappeared without a response
19 SP_DistractorDisappear The currently active distractor has disappeared

Data can be epoched by the stimulus events SP_CueChange, SP_CentralStimulusSpawn, SP_PeripheralStimulusSpawn, SP_DistractorSpawn. ICA can be run either separately on each of these classes of epochs, or on all pooled, or on grouped subsets (for example pooling SP_PeripheralStimulusSpawn with SP_DistractorSpawn). Epochs should be long enough to include the associated behavioural response events.

ERPs and ERSPs can be computed to each of these stimulus events. Difference waves between SP_PeripheralStimulusSpawn and SP_DistractorSpawn may also be of interest. In addition, response errors (see above for definitions) can be analysed for the error-related negativity.

(Originally the intent was to flicker the backgrounds in the four sectors at different fundamental frequencies, so that covert attention to a particular sector would be encoded in the EEG frequency spectrum via a slight increase in SSVEP amplitude at that fundamental frequency. However, many subjects found the flickering distracting, and in addition some autism-spectrum subjects found it painful. The flickering therefore was disabled for most subjects.)