A replication of Marken's Cause ...

RM3-RPT3.DOC

                                 A REPORT

                                     by

                                Chuck Tucker

                     with comments and programming by

                                Rick Marken

        This is a report of a replication (with extention) of Rick
        Marken's study "The Cause of Control Movements in a Tracking
        Task" originally published in PERCEPTUAL AND MOTOR SKILLS,
        1980, 51, 755-758 (Marken Reader Pp. 61-66).

        The task performed is a typical compensatory tracking
        task with several exceptions. Each trial done by a person is
        two "runs" using exactly the SAME disturbance type, pattern
        and speed for each "run." During each "run" the person is
        asked to keep the tip of the moving arrow (cursor) aligned
        with the tip of the stationary arrow at the center of the
        monitor screen. The correlations for these two "runs" are
        the relationship between the arrow (cursor) movements and
        the relationship between the mouse movements.

        A typical Stimulus-Response (S-R) characterization of this
        experience would be that the mouse movements are caused
        by the "stimulus properties" or "intrinsic qualities" of
        the cursor. This being the case, the prediction, would
        be that if the exact same response (R) happens on two
        occasions then the exact same stimulus (S) should have
        happened on those two occasions. In this exercise, if
        the person moves the mouse (R) in exactly the same way
        on two occasions (i.e., "runs") then the cursor movements
        (S) should be the same on the two occasions(i.e., "runs").
        So using the S-R formulation and specification for the
        "stimulus" and "response," the "sameness" for the S and R
        on the two occasions is measured by the correlations
        between the mouse and the cursor movements on those
        occasions. The correlations should be very similar on
        the two occasions if the S-R formulation is correct.

        STUDY PROCEDURES

        Each person (other than myself) doing his exercise
        was a student volunteering from a course in sociological
        social psychology without pay in money or grade points.
        Each one was seated in front of a MAC IIci and asked to
        follow my instructions. All were experienced at using
        a mouse (not this particular mouse on this computer) but
        none had done this exercise or any task similar to it.
        None were allowed to "practice" the task but simply shown
        what would happen on the monitor screen and what numerals
        to record on a form.

        All saw how a "model" looked (a moving arrow) on the screen
        and told that the main difference between watching the "model"
        and their activities would be that they would be aligning the
        arrows by using the mouse while the "model" was doing it from
        a program. During the running of the "model" the participant
        moved the mouse to demonstrate that the mouse had no "effect"
        on the arrow movement. When the "model" finished, two sets of
        lines appeared on the screen with a numeral for each set and
        numerals labelled "First Trial Control," "Second Trial Control,"
        and "Disturbance ID." (SEE BELOW) I showed each participant

        NAME OF EXERCISE: ALIGN 4 DATA DATE: Spring 1994
        DISTURBANCE TYPE: Sinus LEVEL: Very Easy

        TRIAL Cursor-R Mouse-R TC#1 TC#2 D-ID#
        MODEL .405 .999 1.73 1.82 296

        where to record the numerals from the monitor screen on a sheet
        of paper.

        Each trial was two "runs" of forty seconds. After the first
        "run" a question appeared on the screen asking if they
        wanted to do a second "run" and they were told to "click"
        on the box next to the question. After the second "run"
        they were asked if they wanted an analysis and they were
        told to "click" on that box. After 30 seconds a diagram of
        lines and numerals appeared on the screen. They recored
        the numerals from the screen to a form. No one reported
        any difficulties following these instructions and recording
        the numerals.

        DATA

        I report the data for seven trials of two persons done
        under different disturbance patterns. The numerals for
        each trial are: Cursor-R (a correlation between the cursor
        movement on the first "run" and the second "run"); Mouse-R
        (a correlation between the mouse movements on the first "run"
        and the second "run"); MN#1 (a numeral representing the
        amount of "control" on the first "run" [average number of
        pixels the arrow deviated during the "run" - perfect control
        is zero]); MN#2 (a number representing the amount of "control"
        on the second "run" [average number of pixels the arrow
        deviated during the "run" - perfect control is zero]); D-ID#
        (a number used to identify the disturbance for both of the
        "runs"). [The original numerals for TC#1 and TC#2 are "squared
        deviations" while MN#1 and MN#2 are simply the square root of
        those numerals]

        NAME OF EXERCISE: ALIGN 4 DATA DATE: Spring 1994
        DISTURBANCE TYPE: Sinus LEVEL: Very Easy

        PARTICIPANT NUMBER: 001

        TRIAL Cursor-R Mouse-R MN#1 MN#2 D-ID#

          1 .225 .998 1.82 1.68 148
          2 .327 .999 1.62 1.62 254
          3 .244 .998 1.55 2.21 182
          4 .248 .999 1.65 1.58 247
          5 .409 .999 1.24 1.37 105
          6 .444 .999 1.34 1.28 369
          7 .311 .999 1.64 1.56 143

        PARTICIPANT NUMBER: 002

        TRIAL Cursor-R Mouse-R MN#1 MN#2 D-ID#

          1 .433 .999 1.91 1.88 179
          2 .466 .999 1.88 1.64 182
          3 .449 .998 1.86 1.85 337
          4 .360 .999 1.64 1.75 151
          5 .292 .999 1.71 1.88 291
          6 .375 .999 1.81 2.02 263
          7 .355 .999 1.73 1.92 187

···

***************************

        NAME OF EXERCISE: ALIGN 4 DATA DATE: Spring 1994
        DISTURBANCE TYPE: Sinus LEVEL: Easy

        PARTICIPANT NUMBER: 003

        TRIAL Cursor-R Mouse-R MN#1 MN#2 D-ID#

          1 .452 .998 2.25 2.25 336
          2 .407 .996 2.48 3.46 339
          3 .518 .998 2.00 2.51 212
          4 .507 .997 2.63 3.08 188
          5 .511 .998 1.74 2.35 383
          6 .433 .998 2.38 2.09 207
          7 .459 .998 2.52 2.90 224

        PARTICIPANT NUMBER: 004

        TRIAL Cursor-R Mouse-R MN#1 MN#2 D-ID#

          1 .312 .998 2.11 2.15 361
          2 .204 .998 2.10 2.69 179
          3 .228 .998 2.28 2.00 262
          4 .457 .998 1.77 1.99 102
          5 .385 .999 2.23 2.12 249
          6 .498 .998 1.76 2.18 116
          7 .407 .998 2.27 1.89 222

                        ***************************

        NAME OF EXERCISE: ALIGN 4 DATA DATE: Spring 1994
        DISTURBANCE TYPE: Random LEVEL: Very Easy

        PARTICIPANT NUMBER: 005

        TRIAL Cursor-R Mouse-R MN#1 MN#2 D-ID#

          1 .650 .998 1.90 1.98 275
          2 .558 .998 2.18 1.92 295
          3 .519 .998 1.99 1.91 351
          4 .524 .998 1.97 1.92 181
          5 .665 .997 2.06 2.01 373
          6 .568 .998 1.80 1.71 148
          7 .452 .997 1.86 1.95 359

        PARTICIPANT NUMBER: 006

        TRIAL Cursor-R Mouse-R MN#1 MN#2 D-ID#

          1 .465 .997 2.08 2.38 301
          2 .392 .997 2.34 2.61 204
          3 .511 .994 2.54 2.82 377
          4 .537 .997 2.48 2.16 222
          5 .462 .998 1.87 2.08 160
          6 .613 .998 1.98 2.10 177
          7 .560 .998 2.18 2.11 233

                        ***************************

        NAME OF EXERCISE: ALIGN 4 DATA DATE: Spring 1994
        DISTURBANCE TYPE: Random LEVEL: Easy

        PARTICIPANT NUMBER: 007

        TRIAL Cursor-R Mouse-R MN#1 MN#2 D-ID#

          1 .723 .997 3.35 3.24 142
          2 .681 .995 3.60 4.14 219
          3 .603 .994 4.10 3.72 240
          4 .457 .994 2.98 3.30 208
          5 .533 .994 2.63 2.69 354
          6 .788 .998 3.20 3.22 152
          7 .532 .994 2.64 2.65 357

        PARTICIPANT NUMBER: 008

        TRIAL Cursor-R Mouse-R MN#1 MN#2 D-ID#

          1 .745 .996 3.25 3.93 199
          2 .746 .997 3.59 3.15 126
          3 .748 .996 3.27 3.10 368
          4 .780 .996 3.26 3.89 190
          5 .722 .994 3.97 4.36 175
          6 .693 .993 3.36 3.32 363
          7 .797 .996 4.17 3.98 209

                        ***************************

        These data do not seem to support the S-R predictions. The
        variation of the correlations of the mouse movements is
        extremely small with an average correlation of about .995.
        There is much more variation in the cursor correlations with
        an average of about .50 with the smallest r=.204 and the
        largest r=.794. Learning (if it is indicated by more
        "control" in terms of smaller mean pixels) does not seem to
        consistently occur on the second "run" of each trial. There
        appears to be some difficulty in "controlling" across "conditions"
        in that it was best for those in the "sinus, very easy condition"
        while poorest for those in the "random, easy condition" but
        since a different person performed these acts it is not
        possible to readily compare across "conditions." Even though
        the "control" was quite poor for persons #7 and #8 the
        difference between the correlations is still evident.
        It appears, counter-intuitively, that the data for those
        with the best "control" are least likely to be explained
        by the S-R formulation. Perhaps that intuition is derived
        from the S-R formulation rather than the PCT model.