PCT informed silent reading research

CSGnet Archive CSG2001
1

[From Dan Palmer (2000.05.06)]

Hi all,

As part of my PhD research (my guiding question is: "What is psychology
about?") here in Australia, I'm about to conduct a series of
PCT-informed experiments on silent reading. In this post, I was hoping
to get some PCT informed opinions on whether I'm going about things the
right way.

In Serial Visual Presentation (SVP) reading research, successive units
of text (usually words) are presented to a reader at a static location
on a screen. So only one word is visible at any one time: [The] - [cat]
- [sat] and so on, with each new word replacing its predecessor. The
main variable of interest is word presentation rate (WPR), which is
usually either preset by the experimenter, or controlled by the
participant (e.g., depressing one key to increase the rate by 10% and
another to decrease it by 10%).

In my first experiment, taking the basic idea straight from my
understanding of PCT (thanks!), I'm playing both off against each other
and doing a test for rate as a controlled variable. If the reader
touches nothing on the keyboard, WPR speeds up until it is going so fast
as to be unreadable, and then slows down till a value at which it is
going so slow it's damned annoying. At the moment, I have it set so
each speed up-slow down cycle takes about 60 seconds. The participant
is free to control against these disturbances by depressing one key to
increase WPR and another to decrease it. We have the software
(futureBasic 3) up and running, and in preliminary tests the participant
controls the rate beautifully, with lovely feedback patterns emerging in
real-time. WPR a controlled variable? - you betcha!

Well, that describes the basic setup. I wonder if anyone knows of
anything that's already been done along these lines, and whether
presetting the disturbance to vary rhythmically (I was thinking of maybe
exploring a sine-wave pattern so that the rate of key depressions
required to maintain a constant rate is constantly changing) is a
sensible first step (with random disturbance the rate may well stay
within comfort thresholds). Any other advice would be appreciated
before I jump in and start recruiting my first real participants...

Regards,
Dan Palmer

2

[From Bruce Abbott (2001.05.06.1315 EST)]

Dan Palmer (2000.05.06) --

Hi Dan,

As part of my PhD research (my guiding question is: "What is psychology
about?") here in Australia, I'm about to conduct a series of
PCT-informed experiments on silent reading. In this post, I was hoping
to get some PCT informed opinions on whether I'm going about things the
right way.

I hope I can be of service, but I do have a few questions of my own, if you
don't mind. First, I'm unclear about this: If the "guiding question" of
your research is "what is psychology about?", (1) what DO you think
psychology is about, and (2) how does this question lead you into doing
research on reading speed?

In Serial Visual Presentation (SVP) reading research, successive units
of text (usually words) are presented to a reader at a static location
on a screen. So only one word is visible at any one time: [The] - [cat]
- [sat] and so on, with each new word replacing its predecessor. The
main variable of interest is word presentation rate (WPR), which is
usually either preset by the experimenter, or controlled by the
participant (e.g., depressing one key to increase the rate by 10% and
another to decrease it by 10%).

O.K., so WRP is either an independent variable or a controlled variable in
this sort of research.

In my first experiment, taking the basic idea straight from my
understanding of PCT (thanks!), I'm playing both off against each other
and doing a test for rate as a controlled variable.

I don't follow: what are the "both" and how does testing rate as a CV play
one off against the other?

If the reader
touches nothing on the keyboard, WPR speeds up until it is going so fast
as to be unreadable, and then slows down till a value at which it is
going so slow it's damned annoying. At the moment, I have it set so
each speed up-slow down cycle takes about 60 seconds.

What is the pattern of this disturbance? Is it a regular or irregular pattern?

The participant
is free to control against these disturbances by depressing one key to
increase WPR and another to decrease it.

That will work -- in fact a procedure like this is used with rats and
pigeons in operant studies: it's called a "titration" schedule. However, if
I were doing this research I'd probably opt for having participants
manipulate a mouse or joystick to control the WRP rather than having them
repeatedly press keys. You can get a much better time resolution (more data
points/second) and I would suspect that your participants would be smoother
at controlling mouse or joystick motion than keypress rate.

We have the software
(futureBasic 3) up and running, and in preliminary tests the participant
controls the rate beautifully, with lovely feedback patterns emerging in
real-time. WPR a controlled variable? - you betcha!

Well, it's always nice to see the control process working, but of course
it's no surprise that your participants control the word presentation rate.
(I presume that this is just a first step toward answering other research
questions.) Didn't you say at the beginning that giving participants
control of WPR is a standard procedure in this research area? So, what do
you hope to demonstrate that itn't already known?

One thing I'd add to this study if you don't already have it is a measure of
each participant's reading rate under natural conditions. I have a feeling
that under natural reading conditions the participant varies his or her rate
as needed for comprehension. More difficult passages would be read more
slowly than easier ones.

Well, that describes the basic setup. I wonder if anyone knows of
anything that's already been done along these lines, and whether
presetting the disturbance to vary rhythmically (I was thinking of maybe
exploring a sine-wave pattern so that the rate of key depressions
required to maintain a constant rate is constantly changing) is a
sensible first step (with random disturbance the rate may well stay
within comfort thresholds). Any other advice would be appreciated
before I jump in and start recruiting my first real participants...

Rhythmic (periodic) variation of disturbance values introduces some
complications that you might wish to avoid -- there is the possibility that
your participants will learn the pattern and attempt to anticipate the
changes. The presence of such a strategy could complicate the analysis.
Using smoothly varying random disturbances prevents this.

How are you going about analyzing your data?

Best wishes,

Bruce A.

3

[From Dan Palmer (2001.05.08.0950)]

[From Bruce Abbott (2001.05.06.1315 EST)]

Bruce wrote:
I hope I can be of service, but I do have a few questions of my own, if
you don't mind. First, I'm unclear about this: If the "guiding
question" of your research is "what is psychology about?", (1) what DO
you think psychology is about, and (2) how does this question lead you
into doing research on reading speed?

Dan replies:
Hey there Bruce. The Phd is taking the following shape. First, it
looks at how sciences can and have obtained progressively clearer
specifications of their concrete (ostensively definable) basic units
with a cyclic iteration between pointing, classification, and
specification. Second, it reviews most psychology's appalling progress
in this regard. Third, from among those thinkers that _have_ made solid
progress, it reviews the units of Kantor (behavior segment), Skinner
(operant), Lee (deed), and Powers (CS) and suggests a preliminary
integration. The proposed integration ties in Kantor's notion of
multiple participation, Skinner's notion of contingency, Lee's notion of
a change contributed to by the efforts of at least one organism, and
Power's circular characterization of the relations among such changes
(plus disturbances). It is a _preliminary_ specification firmly
grounded in concrete units that are available to the anonymous observer.
One of my main aims in doing this is to explore the means by which
psychologists might start using a shared criterion to assess and
integrate different (and always provisional) theories about the subject
matter. Fourth, it will evaluate this proposed integration in the
context of some data, which are coming from the experiment I described
in my earlier message.

I hope that clarifies how I got from the question "What is psychology
about?" to this experiment. In terms of what psychology is about, my
present take is that it has to do with certain organizations (negative
feedback included) between things done, in the sense of deeds, acts,
outcomes, achievements, accomplishments, or results, where these terms
refer to the satisfying of some well-defined completion criterion (e.g.,
change in state of button from up to down). I'm pretty open minded at
this stage, and am trying to let the data lead me forward.

Dan wrote:

>In my first experiment, taking the basic idea straight from my
>understanding of PCT (thanks!), I'm playing both off against each
other
>and doing a test for rate as a controlled variable.

Bruce queried:
I don't follow: what are the "both" and how does testing rate as a CV
play one off against the other?

Dan replies:
Well, the preset triangular wave disturbances to rate interact with the
participant induced rate changes to determine the rate at any instant.
They play off against each other in the sense that they push rate in
opposing directions. However, I was using the term "play-off" loosely
and don't intend it to replace a more rigorous description of what is
going on.

Bruce queried:

What is the pattern of this disturbance? Is it a regular or irregular
pattern?

Dan replies:
At present it is a regular triangular wave function disturbance with a
cycle length of 60 seconds. However, I'm open to suggestion!

Bruce suggested:
If I were doing this research I'd probably opt for having participants
manipulate a mouse or joystick to control the WRP rather than having
them repeatedly press keys. You can get a much better time resolution
(more data points/second) and I would suspect that your participants
would be smoother at controlling mouse or joystick motion than keypress rate.

Dan replies:
Excellent suggestion! I do plan to try these options later down the
track. Another bonus with a joystick would be the force-proportional
resistance tension. It would sure be a lot smoother. These options
have been explored occasionally in the literature, but I think we'll
stick with key presses for the initial experiment. I find it easier to
start thinking about the relations among events (as a start) when they
are already discrete, the present resolution is plenty enough to keep me
busy, using key-presses is more consistent with the literature we're
working out of, and it's working very smoothly as it is. Tight control
doesn't require a particularly high rate of key depressions. One other
option would be changing the rate every X time the key remains down I suppose...

Bruce commented:
Well, it's always nice to see the control process working, but of course
it's no surprise that your participants control the word presentation
rate. (I presume that this is just a first step toward answering other
research questions.) Didn't you say at the beginning that giving
participants control of WPR is a standard procedure in this research
area? So, what do you hope to demonstrate that isn't already known?

Dan:
I like it! Now, let me try and defend my proposed experiment.
Participants have only ever been given control of undisturbed rate.
What my data will show (if they follow the test-data) is the
'comfort-zone' in which rate is maintained in real time. Existing
research goes about such questions by averaging across the average rates
of participants in different group conditions. What I'm aiming for is a
fast track, data-intensive procedure for obtaining such information from
single participants in single sessions. Further, this kind of data
calls into question the traditional information-processing approach to
it and might help demonstrate why PCT is a much more sensible option in
terms of making sense of the data in all its resplendent complexity.
Finally, of course, the underlying aim is to collect some data rich and
free-flowing enough to use to try and destroy (or preferably refine) my
proposed specification of a concrete psychological unit. Believe me,
this literature is so data-lean and wild-speculation rich that there is
a _lot_ that isn't already known! (like exactly what lurks beneath all
those means of means, for example).

Bruce commented:
One thing I'd add to this study if you don't already have it is a
measure of each participant's reading rate under natural conditions. I
have a feeling that under natural reading conditions the participant
varies his or her rate as needed for comprehension. More difficult
passages would be read more slowly than easier ones.

Dan replies:
Great idea - I should time them reading in the normal format so I can
compare the SVP data. Thanks! Later I hope to manipulate text
difficulty and also the aim of the task (e.g., to count instances of a a
particular word, to summarize the text, or to sit an extensive test on it).

Bruce comments:
Rhythmic (periodic) variation of disturbance values introduces some
complications that you might wish to avoid -- there is the possibility
that your participants will learn the pattern and attempt to anticipate
the changes. The presence of such a strategy could complicate the
analysis. Using smoothly varying random disturbances prevents this.

How are you going about analyzing your data?

Dan replies:
That is another good point, thanks. I will explore the notion of
smooth-random disturbance curves, perhaps after we run a participant to
see what happens over a larger time frame. Where might I find some info
on programming smooth random disturbance? One possible benefit of the
regularity is that I can have the uncontrolled for preset rate
disturbance function as a background on the graphs which makes them a
lot more comprehensible.

As for data analysis, I'm using scientific visualization software
(Matlab) to try and see patterns among the recorded events in a
condensed visual format. A useful plot so far has been rate over time
with superimposed different symbols representing the different key
events. One thing I'd like to explore is using logical testing to pull
out more order. An example would be if preset rate is incrementing and
if rate > 400 word per minute, how likely is a slow down key depression?
This kind of approach has been used successfully in the past by my
supervisor Vicki Lee. I also want to try and model the data PCT style.
I figure that is probably going to teach me more about PCT than just
reading the books!

Thanks again to you Bruce for your stimulating and much-welcomed comments,
Dan

4

[From Bruce Abbott (2001.05.08.2120 EST)]

Dan Palmer (2001.05.08.0950) --

That is another good point, thanks. I will explore the notion of
smooth-random disturbance curves, perhaps after we run a participant to
see what happens over a larger time frame. Where might I find some info
on programming smooth random disturbance? One possible benefit of the
regularity is that I can have the uncontrolled for preset rate
disturbance function as a background on the graphs which makes them a
lot more comprehensible.

You can do the same with a random disturbance, if you have your program save
the disturbance values into an array. Below is a snippet of Pascal code
that illustrates how to create a smoothly-varying random disturbance with a
specified maximum value. (I borrowed it from one of Bill Powers' programs.)
I present some code and then comment on it. Pascal code is easy to read
even if you are not familiar with it; you should be able to translate it
fairly easily to whatever programming language you are using.

···

-------------------------------------------------------------------------
const
  MAXDATA =1800;

{ The line above defines the constant MAXDATA, which is the array size. }
{ Here I've set it to 1800, the number of disturbance values you would }
{ need if you were running at 30 data points per second for 60 seconds }
{(30 * 60 = 1800). }

var
  d: array[1..MAXDATA] of integer;

{ This line sets up an array to hold the 1800 integer disturbance values. }
{ The disturbance values are then created in procedure InitDist: }

procedure InitDist;
var i, max: integer;
    slow, d1, d2, d3, avg, tmp: real;

begin
  randomize;
  { This sets the random number generator to produce a new sequence each }
    time the program runs. }
  slow := 0.005; { rapidity of disturbance, smaller = slower }
  d1 := 0.0; d2 := 0.0; d3 := 0.0;
  for i := 1 to MAXDATA do
    { Create the raw disturbance values and store in array d[i] }
    begin
      d1 := random * 10000.0 - 5000.0;
       { Random is a number between 0 and 1 from the random number }
         generator. The next two steps do the smoothing using "leaky }
         integrators." }
      d2 := d2 + slow*(d1 - d2);
      d3 := d3 + slow*(d2 - d3);
       { The real number d3 is then converted to an integer value by }
         rounding and stored in d[i]. }
      d[i] := round(d3);
     end;
  { What follows below normalizes the disturbance values to have an average }
    deviation of zero: }
  avg := 0.0;
  for i := 1 to MAXDATA do avg := avg + d[i];
  avg := avg/MAXDATA;
  for i := 1 to MAXDATA do d[i] := d[i] - round(avg); }
  { Finally, the largest absolute deviation is found and the disturbance }
    values normalized to a maximum absolute value of 120. Change this }
    to whatever value is needed in your application. }
  max := 0;
  for i := 1 to MAXDATA do if abs(d[i]) > max then max := abs(d[i]);
  for i := 1 to MAXDATA do { normalize to max of 120 }
    begin
      tmp := 120.0/max;
      d[i] := round(d[i]*tmp);
    end;
end;
----------------------------------------------------------------------

If you have your data-collection program save both the disturbance values
and your data values at each time increment, you can plot both just as you
are now doing with your triagnular disturbance pattern.

Hope this helps!

Bruce A.

5

For some reason this note to Dan Palmer kept bouncing ("too many hops"
was the complaint, which seems appropriate since Dan lives down there in
Kangaroo land) so I am posting it to CSGNet.

Hi Dan --

Exactly - although in my preliminary trials it is actually quite hard to
keep word duration perfectly straight - for me it still varied within
about 20 ticks (333.333 ms) as you can see (actually, however, I wasn't
controlling so much for rate as for my ability to read the text, which
is interesting...). The effect is still very clear, however.

I was able to open your plots this morning (because I have the correct
version of WORD here at work) and they are gorgeous!! Don't worry about
keeping the word duration (or inverse of word duration) completely
straight; you'll never to that.

The plot if inverse word duration (p) is really the best. If you add the
disturbance (d) and integrated value of the keypress results (o) to the
plot, you will see the nice mirror image relationship between o and d
and the relatively flat plot of p. What you should do now is _measure_
control. I would use the stability measure which is:

S = 1- sqrt (expected variance of p)/(observed variance p).

The observed variance of p is just a measure of the variance (over time)
of p (inverse word duration). The expected variance of p is the sum of
the variances of d and o: var(d)+var(o). The more negative the S
(stability) measure, the better the control of word duration. So a
stability value of -10 is _very_ good.

Using S as the measure of control you can test hypotheses about other
possible controlled variables. For example, since word duration is
confounded with word onset time difference you might try a run where
word duration is fixed at some minimum value and let the person control
time between word onsets. See if control of this variable is any better
(bigger S value) than control of word duration.

You can use S to hone in on the best definition of the controlled
variable. There are many possibilities, even in your apparently simple
situation. Maybe average word length has something to do with it; the
controlled variable might be duration/average word length. This
experiment could be a great way to show behavioral scientists how to
test for some pretty interesting controlled variables.

Anyway, the plots of just the simple, basic control of duration results
are very interesting in themselves. You have certainly shown that
people can control word presentation rate (or a variable much like it).
That, in itself, would be huge news to the behavioral science community,
which believes that word duration controls behavior.

Best regards

Rick

···

--
Richard S. Marken, Ph.D.
RAND M-23
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

6

[From Bruce Abbott (2001.05.09.1150 EST)]

This is directed mainly to Dan Palmer, and was occasioned by Rick Marken's
post to Dan on this topic on CSgnet today. It represents an attempt to
correct what I perceive as a misleading statement in Rick's post.

Rick Marken

You have certainly shown that
people can control word presentation rate (or a variable much like it).
That, in itself, would be huge news to the behavioral science community,
which believes that word duration controls behavior.

I wouldn't take this last sentence too seriously, Dan -- it misleads by
failing to take acount of the fact that the word "control" as used by
behavioral scientists and many others usually refers to something other than
"control" as used in this forum. In this forum it means a process whereby
an input variable is brought toward and maintained near some reference value
by taking action as needed. In that other context, it means that changes in
one variable produce changes in another variable, when other variables are
held reasonably constant. In other words, to say that A "controls" B is to
say that A influences B. Word duration (length?) "controls" behavior in
this sense: when you use different word durations (lengths?), you get
different behavior -- different reading rates in this case. This would
follow naturally from the fact that longer words take longer to read.

Rick asserts that psychologists would be surprised to learn that people
control (in the first sense) their reading rate or some variable "much like
it." I wouldn't take that claim too seriously either. The rate at which a
person reads may vary with numerous factors (e.g., familiarity of the words,
word length, difficulty of the sentences, complexity of the ideas being
presented, and so on) but clearly a person will read at some rate that he or
she finds comfortable and permits what the person perceives to be an
adequate level of understanding of the material. I rather doubt that this
has escaped the notice of those doing research on reading.

I agree that what you are doing is the best way to gain a good understanding
of the control process. However, from what you've said it sounds to me as
though you have not yet conducted a thorough review of the existing
literature in this area. I'd certainly want to do this before investing too
much time and effort on this project, unless your goal is purely one of
using this as a test bed for learning how to develop and test control
models. I wouldn't be surprised to find that your project or something very
much like it has already been done.

Best wishes,

Bruce A.

7

i.kurtzer (2001.05.10.1200)

Bruce, why don't you stop pissing on someone's attempt to do research.

i.

···

----- Original Message -----
From: Abbott_Bruce <ABBOTT@IPFW.EDU>
To: <CSGNET@LISTSERV.UIUC.EDU>
Sent: Wednesday, May 09, 2001 11:53 PM
Subject: Re: PCT informed silent reading research

[From Bruce Abbott (2001.05.09.1150 EST)]

This is directed mainly to Dan Palmer, and was occasioned by Rick Marken's
post to Dan on this topic on CSgnet today. It represents an attempt to
correct what I perceive as a misleading statement in Rick's post.

>Rick Marken

>You have certainly shown that
>people can control word presentation rate (or a variable much like it).
>That, in itself, would be huge news to the behavioral science community,
>which believes that word duration controls behavior.

I wouldn't take this last sentence too seriously, Dan -- it misleads by
failing to take acount of the fact that the word "control" as used by
behavioral scientists and many others usually refers to something other

than

"control" as used in this forum. In this forum it means a process whereby
an input variable is brought toward and maintained near some reference

value

by taking action as needed. In that other context, it means that changes

in

one variable produce changes in another variable, when other variables are
held reasonably constant. In other words, to say that A "controls" B is

to

say that A influences B. Word duration (length?) "controls" behavior in
this sense: when you use different word durations (lengths?), you get
different behavior -- different reading rates in this case. This would
follow naturally from the fact that longer words take longer to read.

Rick asserts that psychologists would be surprised to learn that people
control (in the first sense) their reading rate or some variable "much

like

it." I wouldn't take that claim too seriously either. The rate at which

a

person reads may vary with numerous factors (e.g., familiarity of the

words,

word length, difficulty of the sentences, complexity of the ideas being
presented, and so on) but clearly a person will read at some rate that he

or

she finds comfortable and permits what the person perceives to be an
adequate level of understanding of the material. I rather doubt that this
has escaped the notice of those doing research on reading.

I agree that what you are doing is the best way to gain a good

understanding

of the control process. However, from what you've said it sounds to me as
though you have not yet conducted a thorough review of the existing
literature in this area. I'd certainly want to do this before investing

too

much time and effort on this project, unless your goal is purely one of
using this as a test bed for learning how to develop and test control
models. I wouldn't be surprised to find that your project or something

very

much like it has already been done.

Best wishes,

Bruce A.

8

[From Bruce Abbott (2001.05.1225 EST)]

i.kurtzer (2001.05.10.1200) --

Bruce, why don't you stop pissing on someone's attempt to do research.

Hi Isaac. I'm sorry you perceive it that way -- and I'm a bit taken aback
that you do. I hope and believe that Dan sees my comments as constructive,
not destructive, criticism. I think it's great that Dan is applying control
theory in this area.

It was certainly my intention to be constructive. Thus far I've (1)
suggested using what may be a better means of control than rate of
keypressing, (2) provided the method for creating a slowly-varying random
disturbance that Dan asked me for, and (3) suggested that a thorough review
of the literature in this area might be in order before progressing much
further with designing experiments. Which of these do you perceive
constitutes "pissing"?

By the way, I think it would be nice if you would describe some of your own
research here on CSGnet. How is it going? What results have you obtained?

Best wishes,

Bruce A.

9

Isaac,

I most strenuously OBJECT to your sophomoric attempts at controlling
behavior! Either communicate your displeasure off-net with the person you
are vexed at, or be a mensch and please keep it professional.

Regards,

Bryan Thalhammer

···

i.kurtzer (2001.05.10.1200)

Bruce, why don't you stop pissing on someone's attempt to do research.

i.

10

i.kurtzer (2001.05.10.1300)

Dan, I think you are making a good first start.
I would now introduce the quantitative measures Rick suggests as well as
formulate a number of other
hythetical controlled variables. Word rate [not length--if the word rate is
stable over a two or three fold change in length for the same rate of
length, for example], and syllable rate come to mind, but there are as many
as in your imagination. Good luck.

p.s. how do i remember the name Vicki Lee? You said she's your advisor,
right.

i.

11

[Dan Palmer 2001.5.12.1300]

Hi all,

I'm on daily digest mode and don't seem to have received the post from
Rick that Bruce critiques nor the advice on smoothed random disturbance
that Bruce mentions (though Rick has helped me out and I have SRD up and
running beautifully - thanks Rick!). I'd appreciate it if particularly
the former could be forwarded to me.

****************** Bruce A. wrote:******************
"I agree that what you are doing is the best way to gain a good
understanding of the control process. However, from what you've said it
sounds to me as though you have not yet conducted a thorough review of
the existing literature in this area. I'd certainly want to do this
before investing too much time and effort on this project, unless your
goal is purely one of using this as a test bed for learning how to
develop and test control models. I wouldn't be surprised to find that
your project or something very much like it has already been done"

****************** Dan replies ******************
Hi Bruce. I'm afraid I'm not sure where you are coming from. Over the
last year and a half I have collected, read, and reviewed hundreds of
articles - virtually the entire serial-visual-presentation reading
literature (which Forster originated here at Monash in 1971) in what
seems to me to be a fairly extensive manner. I have also gone back
through the early operant work on this kind of interface, which was
pioneered in an clever little study by Goldiamond in 1962 (which in turn
grew out of Holland's earlier work on what he called "observing
responses"). I have conducted and submitted an earlier experiment
(without the PCT component) in this literature, and my supervisor and I
are currently writing up a review paper of this very literature for
submission. I can assure you with some confidence that nothing even
closely resembling my planned experiment has ever been conducted before.
I'm sure you can appreciate that in this context your above comments
come across as rather bizarre!

Most of the existing SVP studies have preset word presentation rate and
forced readers to simply cope with it, on the grounds of testing various
hypothesis about inner processing mechanisms. While a few studies have
explored enabling the reader to influence word presentation rate in real
time (e.g., a key depression for each word, sentence, or to change the
rate by 10%), none have disturbed rate to enable an assessment of the
ongoing dynamics of reader control (in the PCT sense). It is even very
rare for SVP studies (the operant work excluded - Goldiamond, for
example, presented cumulative records) to even present any data directly
corresponding to what participants were actually doing in realtime (in
many of the studies they may well have been looking out the window!).

I think it a great shame that you don't appreciate the potential value
of such a contribution in its own right, or at least provide some solid
rational or empirical grounds for your skepticism. That said, I again
thank you for your solid procedural recommendations which I do appreciate.

****************** i.kurtzer (2001.05.10.1300) wrote:******************

Dan, I think you are making a good first start. I would now introduce
the quantitative measures Rick suggests as well as formulate a number of
other hypothetical controlled variables. Word rate [not length--if the
word rate is stable over a two or three fold change in length for the
same rate of length, for example], and syllable rate come to mind, but
there are as many as in your imagination. Good luck.

p.s. how do i remember the name Vicki Lee? You said she's your advisor, right.

****************** Dan Replies ******************
Hey Isaac,

Thanks for your kind words. There are certainly a plethora of
controlled variables to hypothesize about (recall I haven't yet seen the
full post of Rick's you refer to - but look forward to it - Rick has
already shared some very exciting suggestions privately) - a million
directions the research could go in (e.g., sentence rate, within
sentence rate, word duration, word rate, pause size between words, even
contrast, fontsize etc...). Your syllable rate suggestion is an
interesting and welcome addition. In practice, word rate is a little
tricky given that it is not linearly related to word duration (If I've
figured this right, word duration in seconds = 60/Words per minute _if_
that duration is held constant). I find it a bit weird to think about a
word presentation rate that changes for every word, for example, because
"rate" is meaningless with reference to just one event. So I'm still
deciding whether to run with rate or duration disturbances - any advice
appreciated! Now, back to programming! I'm keen to produce my first
(smoothed) disturbances vs opposing actions on top of the (hypothesized)
controlled variable, all across time! (The famous PCT mirror-plot!)(I'm
going a little crazy with the exclamation marks, I know, but to hell
with it! - It's a beautiful day, and I'm in a great mood!)

Oh yeah - Vicki Lee is my supervisor, you might have come across her
work. She's been publishing conceptual critiques of traditional
(cognitive and behavioral) psychology and developing a
conceptual/empirical alternative for the last 25 years or so (she
started out in behavior analysis). See
http://www-personal.monash.edu.au/~vlee/

Best,
Dan

···

-----
http://www-personal.monash.edu.au/~dpalmer/psychology.htm

12

[From Bruce Nevin (01.05.12 09:04 PDT)]

Dan Palmer 2001.5.12.1300 --

Maintaining a steady syllable rate would be equivalent to slowing word rate for more complex words. Another measure of word complexity is morpheme rate. Other things that slow reading are word familiarity, syntactic complexity, discourse coherence, and what might be called argument complexity (an aspect of discourse structure). I don't think it is very feasible (or useful for your purposes) to try to measure these factors directly, but by measuring word or syllable rate as subjects control it you may find that they vary the rate according to whether they can easily skim or must work a bit to put things together.

  Bruce Nevin

13

[From Bruce Abbott (2001.05.13.1625 EST)]

[Dan Palmer 2001.5.12.1300]

I'm on daily digest mode and don't seem to have received the post from
Rick that Bruce critiques nor the advice on smoothed random disturbance
that Bruce mentions (though Rick has helped me out and I have SRD up and
running beautifully - thanks Rick!). I'd appreciate it if particularly
the former could be forwarded to me.

There must be some problem with transmission along the way as both posts you
mention were distributed via CSGnet and you should have received them in the
digest. I would suggest getting on the web and downloading the current
CSGnet log file at

ftp://postoffice.cso.uiuc.edu/csgnet/

The log file you want (still the current one) is csgnet.log0105b

****************** Bruce A. wrote:******************
"I agree that what you are doing is the best way to gain a good
understanding of the control process. However, from what you've said it
sounds to me as though you have not yet conducted a thorough review of
the existing literature in this area. I'd certainly want to do this
before investing too much time and effort on this project, unless your
goal is purely one of using this as a test bed for learning how to
develop and test control models. I wouldn't be surprised to find that
your project or something very much like it has already been done"

****************** Dan replies ******************
Hi Bruce. I'm afraid I'm not sure where you are coming from. Over the
last year and a half I have collected, read, and reviewed hundreds of
articles . . .

. . .I can assure you with some confidence that nothing even
closely resembling my planned experiment has ever been conducted before.
I'm sure you can appreciate that in this context your above comments
come across as rather bizarre!

As you had not mentioned doing this work before, my suggestion to review the
literature before continuing along this line was anything but bizarre, and I
find it terribly disappointing that you would apply that term to it.

I think it a great shame that you don't appreciate the potential value
of such a contribution in its own right, or at least provide some solid
rational or empirical grounds for your skepticism. That said, I again
thank you for your solid procedural recommendations which I do appreciate.

Now you have drawn the conclusion that I "don't appreciate the potential
value of such a contribution." I did say that such a contribution would be
of little value if it had already been done. You evidently missed the IF part.

Listen Dan, you already have competent help from Rick Marken. I was trying
to be helpful (no one else responded to your request for help on CSGnet and
I did not know that you were getting good advice privately) but as it seems
that we are not communicating well I'm going to leave it at that. Good luck
with the research! I do wish you well.

Best wishes,

Bruce A.