Special Features

Special operations

The pushbutton in the first row of controls selects from among several unique operations that are associated with a single designated segment of the trial. These operations implement very specific experimental paradigms that are not otherwise supported by the definition of a Maestro trial. Several of these special operations, but not all, are triggered by or otherwise involve a saccade. The spinner control labelled at seg designates the segment at which the operation begins, and the Sacc VT edit controls lets the user set the saccade threshold velocity in degrees/second; any integer value in [0..999] may be entered here. Maestro "detects" a saccade whenever the subject's horizontal or vertical eye velocity exceeds this threshold; the saccade ends when the eye velocity falls back below the threshold.

If the pushbutton's label reads none, then there is no special operation defined on the trial, and the other two controls are disabled. Click several times on the pushbutton to see the different operations supported:

• • skipOnSacc - Short for "Skip On Saccade", this saccade-triggered feature is enabled only during the designated segment. If a horizontal or vertical saccade occurs during this segment, trial execution immediately skips to the start of the next segment. A brief marker pulse is delivered on DO<6> if and when the saccade is detected; for analysis purposes, it may be useful to record this marker in the trial data file. Implementation constraints: This feature is not supported for trials in which the Chair target participates. It is also incompatible with any of the other special operations, XYScope target interleaving, or velocity stabilization.

  • selByFix - Another saccade-triggered operation, "Select By Fix" was introduced to support a "two-choice" experimental protocol. It requires that two fixation targets be specified during the designated segment and all segments thereafter. Fixation checking is disabled in the designated segment, during which the animal is supposed to "select" one of the targets by making a saccade to it. When a saccade occurs, Maestro determines which fixation target is closest to the eye at the end of the saccade (when eye velocity falls back below threshold). If the closer target is within the fixation window, it is considered "selected", and the unselected target is turned off for the remainder of the trial. If neither target is within the fixation window at the end of the saccade, Maestro will continue attempting to select a target after each saccade until the designated segment ends. If neither target is selected by the segment's end, the target nearest the eye is chosen. Whenever it occurs, target selection is marked by a pulse delivered on DO<6>. Typically, the animal is expected to maintain fixation on the selected target during the rest of the trial. The identity of the target selected determines the length of reward pulse delivered at trial's end: the first reward pulse duration is used if the Fixation Target #1 was selected, else the second reward duration is used. Thus, the researcher could train the animal to preferentially select one target over the other. Implementation constraints: Same as those described for the skipOnSacc feature, plus the aforementioned rules for identifying the two fixation targets.

  • selByFix2 - In this alternate version of selByFix, a fixation target T is selected if, after a saccade, the eye is closer to the current position of T or what its current position would be had it NOT been displaced at the start of the designated segment. In all other respects, the two versions are the same. In fact, if neither fixation target is displaced at the beginning of the designated segment, selByFix2 works exactly the same as the original version. Rarely used. Implementation constraints: Same as those described for selByFix.

  • chooseFix1, chooseFix2 - These two special operations offer a variation on selByFix in which the animal is required to choose (i.e., fixate upon) a "correct" target over an "incorrect" one, rather than freely choose among two possible targets. During the designated segment, fixation checking is disabled until the moment the animal's eye position falls within the fixation window of the correct target. Unlike selByFix, eye position is checked continuously (once per ms) rather than at the end of each detected saccade. For chooseFix1, Fixation Target #1 is the correct target; for chooseFix2, it is Fixation Target #2. If and when the correct target is "chosen", the "incorrect" target is immediately turned off for the remainder of the trial, a reward pulse is delivered, and fixation checking is re-enabled. As in the selByFix paradigm, the animal is typically expected to follow the chosen target for the remainder of the trial. The target choice event is also marked by a pulse on DO<6>. If the correct target is not fixated by the end of the special segment, the trial aborts because the animal has failed to make the correct choice; in this case, the animal gets no reward whatsoever. The first reward pulse length sets the duration of the end-of-trial reward, while the second reward pulse length gives the duration of the "correct target chosen" reward; typically, the end-of-trial reward will be the larger of the two. Implementation constraints: Same as those described for selByFix.

  • switchFix - This operation was introduced to support a forced two-choice protocol: deliberately encourage the animal to switch from one fixation target to the other after the designated segment, while allowing it to select either target as the "initial selection" during that segment. Like the selByFix operation, two fixation targets must be specified for the designated segment and remain unchanged for the remaining trial segments. Unlike selByFix, the operation is not triggered by a saccade, neither target is turned off and the fixation checking rules are different. For the designated segment and all segments thereafter, fixation checking is enforced except during saccades. Furthermore, the fixation requirements are considered satisfied so long as the eye is within the fixation window of EITHER target. Thus, the animal may freely "switch" between the two designated fixation targets. However, the rule for getting a reward at trial's end is unique: To get the reward, the animal must be fixating on the target that it was NOT fixating at the end of the designated segment. In other words, between the end of the designated segment and the end of the trial, the animal must have switched its fixation from one target to the other in order to get a reward. Implementation constraints: Same as those for selByFix; in addition, there must be at least one trial segment following the designated segment.

  • R/P Distro - Trials using this feature participate in a response distribution-based reward/penalty protocol, first introduced in Maestro v1.4.0. The protocol -- which aims to alter the distribution of a subject's behavioral responses to a rote visual task through motivation -- is unique in that it requires Maestro to accumulate response data over repeated presentations of a trial. The "behavioral response" can be one of four quantities, all related to the subject's eye velocity trajectory: horizontal eye velocity in deg/s, vertical eye velocity in deg/s, the vector eye velocity in deg/s, or the direction of eye motion in deg CCW (0 deg = rightward motion); in all cases, the response is averaged over the course of the special segment, which will typically be quite short in duration, on the order of 20 milliseconds. With each trial repetition, Maestro builds up a response sample histogram that characterizes the variability in the subject's response to the stimulus presented during the trial. Once that histogram is "adequate", the researcher pauses trial sequencing, defines and enables a response reward window for the trial, resets the sample histogram, and then resumes trial sequencing. The trial is presented as before, but now the subject receives a reward or "penalty" (really a reduced reward) if the behavioral response falls within or outside the window, respectively; go here for a full discussion of the reward contingencies. Maestro again accumulates the response data and displays a response histogram -- so the researcher can tell immediately whether or not the response variability has been significantly modified. Runtime control of this special protocol is handled by the R/P Distro dialog tab in the Trial mode control panel; see the description of this tab page for more information.

  • searchTask - In this special operation, first introduced in Maestro v2.6.5, the subject "seeks" a goal target among one or more "distractors". The search task takes place during the designated special segment. Normal fixation checking is suspended entirely during that segment. Instead, on a millisecond-to-millisecond basis, Maestro compares the subject's eye position to the current positions of all targets that are turned on for the special segment. The subject is considered to be fixating a target so long as the eye-target distance is within the specified fixation window. If the subject maintains such fixation continuously for a specified period, then that target was successfully "found" and the trial ends immediately. If the goal target was the one found, then reward pulse #1 is delivered; if a distractor was found, reward pulse #2 is delivered (typically, reward pulse 2 will be shorter than pulse 1 so that a smaller liquid reward is delivered). If the subject's eye wanders outside the screen display bounds, or if the end of the special segment is reached and no target was found, the trial ends and NO reward is delivered. Thus, in a searchTask trial, the special segment is always the last segment. The "goal" target is that designated as Fixation Target #1 during the special segment; all other trial targets that are turned on during the segment are "distractors". The required time-on-target to satisfy the search task is given by the segment's Grace Period, while H,V Fixation Accuracy define the bounds of the on-target window. Finally, even if no reward is delivered, the trial is deemed a "completed" trial if the animal attempted the task, i.e., so long as the animal's eye velocity reaches the value specified in the Sacc VT control.

  • searchTask with 2 goal targets - [as of Maestro v4.2.0] In this version of the searchTask feature, both Fixation Target #1 and Fixation Target #2 are designated and turned on during the search segment. The implementation is the same, except that now the subject can "find" either of 2 goal targets or a distractor, and the reward logic is different. If the subject "finds" Fixation Target #1, then reward pulse #1 is delivered; if the subject finds Fixation Target #2, then reward pulse #2 is delivered. No reward is delivered if the subject finds a distractor or fails to find any target. In typical usage, one reward pulse will be significantly longer than the other to "encourage" the subject to select the corresponding goal target.

XYScope Target Interleaving

Apparent motion is the perception of motion when a sequence of static images are presented with the appropriate spatial and temporal relationships. By giving the user control over the XYScope frame period on a segment-by-segment basis, a Maestro trial permits studies of apparent motion using the XY oscilloscope display. [Such studies are not possible on the RMVideo display, because CRT monitor refresh rates are not fast enough.] However, for frame periods much greater than 16 ms, a target on the XYScope may exhibit a pronounced flashing effect because it is rendered at the beginning of each frame and decays (phosphor decay) significantly over the remainder of the frame. XYScope target interleaving was introduced to reduce this flashing effect, which can complicate the interpretation of apparent motion studies. When interleaving is engaged, each segment's XYScope frame period, P, is subdivided into N subframes, or interleave slots, of duration P/N where N is the number of XYScope targets to be interleaved. Each interleaved target is drawn during a different subframe; any remaining non-interleaved targets are rendered in the first subframe. Thus, to reduce the flashing effect in an apparent motion trial, the experiment designer creates a "composite" target made up of N identical targets all following the same trajectory. During the trial, each component target is drawn in a different subframe, effectively spreading out the target's rendering over the entire frame period and thereby reducing the flashing effect. To use interleaving, simply set the number of XYScope targets to be interleaved using the #XY Interleaved spinner control. Implementation constraints: If N is the number of targets to be interleaved, then there must be at least N XYScope targets participating in the trial -- not including the Spot/Rect Dot Array target type, which is exempt from interleaving. Each segment's XYScope frame period P must be a multiple of N and the subframe period P/N must be a multiple of the minimum XYScope frame period (2 ms).

Overriding display settings that control the XYScope dot pattern seed

The random-dot patterns associated with many XYScope targets are generated by a simple pseudorandom number generator in the XYScope controller's firmware. This generator is seeded with an integer value before generating the dot patterns for all extended XYScope targets participating in a trial. The seed value can be set on the Video Display tab on the mode control panel in most operational modes. However, it is also possible to override the display settings on a trial-by-trial basis via the XY dot seed parameter. This numeric edit control accepts any integer value in the range [-1 ... 999999999]. If the value is -1, it is ignored and the Video Display settings provide the seed. If the value is 0, the seed is randomly generated each time the trial is presented (so the target dot patterns will be different for each trial) -- regardless of the Video Display settings. If the value is positive, then that value is used as the fixed seed for every presentation of that trial (so the target dot patterns are exactly the same for each trial).

[NOTE: The XY dot seed parameter does NOT apply to RMVideo targets that involve a random-dot pattern. Instead, the seed value is part of the RMVideo target's definition.]

Mid-trial Rewards

Typically, a single reward is delivered at trial's end as long as all fixation requirements have been met; if fixation is "broken" at any time, the trial is aborted immediately. For longer trials, a single reward may not be adequate to maintain the animal's interest. To address this issue, Maestro allows the user to enable intertrial rewards on a segment by segment basis. The third row of controls in the Features group box are dedicated to this feature. A pushbutton toggles between two different modes: periodic, in which mid-trial rewards are delivered at regular intervals as long as fixation requirements are met; and atSegEnd, in which a reward is delivered at the end of each segment that is enabled for mid-trial rewards. The numeric edit controls intv and len specify in milliseconds, respectively, the reward interval for periodic mode and the reward pulse length. The reward interval is restricted to integral values in the range [100..9999], with a default value of 1000. The pulse length is restricted to integers in [1...999], with a default value of 10. Note that there is the potential problem of delivering a mid-trial reward near the end of the trial, conflicting with the normal end-of-trial reward. Maestro does NOT protect against this situation, except that in atSegEnd mode, a mid-trial reward will never be delivered at the end of the trial's last segment. Users are responsible for designing their trials appropriately.

Staircase sequencing

Staircase sequencing is commonly used in human psychophysical experiments to determine a subject's threshold response to some visual stimulus. Maestro's Trial mode supports staircase sequencing, and the bottom row of parameter controls in the Features group are related to this sequencer mode. Maestro can intertwine up to five independent staircase sets simultaneously (STAIR1 to STAIR5), and the staircase ID -- set by pushbutton -- indicates to which staircase set a trial belongs. If the staircase ID is NORMAL, then the trial does not belong to any staircase but instead is part of an "irrelevant" set of trials that may be occasionally presented to keep the subject "honest". The trial's staircase Strength is an arbitrarily assigned floating-point value that determines the staircase tier, or step, to which the trial belongs; the strength value is ignored for NORMAL trials. It can be any real number in [0..1000), with up to 3 decimal digits. The Response pushbutton indicates which analog input channel should be monitored for a "correct" response to the trial. Staircase trials are defined so that the subject makes a two-choice decision at some point during the trial. Currently, analog input channels AI<12> and AI<13> are dedicated to monitoring the states of two mechanical pushbuttons representing the two choices the subject can make. The experiment designer can set up the trial so that either the AI<12> or AI<13> button should be depressed for the correct response. The designer also specifies, via the segment table, those segments during which the two response channels should be monitored for the correct (or incorrect) response. This infrastructure is essential so that Maestro can sequence through the staircase trials automatically until it "finds" the subject's threshold. [Note that it is rather inefficient to use two analog input channels to monitor the subject's response when, with the appropriate external electronics, two digital inputs should be sufficient. Maestro may be modified at a later date to use digital inputs instead.]