[From Rick Marken (2007.03.05.1430)]
Looks like "mindreading" is in. Check out:
http://www.cnn.com/2007/TECH/science/03/05/mindreaders.ap/index.html
I guess my demonstration of mindreading would be too "simple" for the
media. Nowadays, you gotta have an MRI machine in order to make an
impression;-)
Best
Rick
--
Richard S. Marken
rsmarken@gmail.com
marken@mindreadings.com
[From Fred Nickols (2007.03.05.1800 EST)] --
I saw this same story on another site. It might be wishful thinking on my part but I think it's wishful thinking on their part.
--
Fred Nickols
www.nickols.us
nickols@att.net
-------------- Original message ----------------------
From: Richard Marken <rsmarken@GMAIL.COM>
[From Rick Marken (2007.03.05.1430)]
Looks like "mindreading" is in. Check out:
http://www.cnn.com/2007/TECH/science/03/05/mindreaders.ap/index.html
I guess my demonstration of mindreading would be too "simple" for the
media. Nowadays, you gotta have an MRI machine in order to make an
impression;-)Best
Rick
--
Richard S. Marken
rsmarken@gmail.com
marken@mindreadings.com
[Martin Taylor 2005.05.16]
Anyone who has tried Rick's "mind reading" demo, in which the computer determines which of several objects the person is trying to control, should look at http://www.dcs.gla.ac.uk/~rod/publications/WilMur04-CHI.pdf.
It was passed on to me with the subject heading: "interesting (and vaguely PCT-inspired) control interface".
Martin
[From Rick Marken (2005.05.16.2130)]
Martin Taylor (2005.05.16) --
Anyone who has tried Rick's "mind reading" demo, in which the computer determines which of several objects the person is trying to control, should look at http://www.dcs.gla.ac.uk/~rod/publications/WilMur04-CHI.pdf.
What a great find! I'll have to get in touch with the authors.
It was passed on to me with the subject heading: "interesting (and vaguely PCT-inspired) control interface".
That is hysterical. Finally an actual PCT-based research project and all it gets is "vaguely PCT inspired".
Thanks for the article.
Best
Rick
---
Richard S. Marken
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400
This is Phil Runkel replying to Rick Marken�s (22 May) enthusiasm about the paper by Williamson and Murray-Smith mentioned by Martin Taylor.
I, too, am delighted that Williamson and Murray-Smith undertook their reading, thinking, and experimentation and wrote about about it. I liked their ideas for variations and applications.
At the same time, I wish I understood their paper better. A large part of my difficulty, I am sure stems from their use of technical terms and manner of expression peculiar to experts in computers. Another source of my difficulty arises from the sheer omission of several small pieces of information that would, I think, improve my understanding.
This communication is addressed primarily to Rick Marken. I hope he can answer some of my questions about the article. But I doubt he can answer all of them, and perhaps there are others reading this who can answer some that Rick cannot.
You might wonder why I do not send this to W&M-S. For one thing, I do not have an address for them. For another, I think it likely that members of the CSGnet can explain W&M-S to me more effectively that W&M-S can.
So if you can clear up some of my puzzlement, please do.
On the first page, right column, 8th line from the bottom, the authors say that �... Powers proposed that his viewpoint provides an empirical method for the subject�s intention.� I looked rather quickly into B:CP and the two volumes of collected papers and could find no such statement. I do know that Marken wrote up a couple of experiments on intention. I think W & M-S should have cited Marken. Am I correct?
Page 2, left column, line 7. What is an �agent�? �Agents try to ...� sounds as if agents are people. And the last paragraph in that column says, �Viewed from an agent perspective...� That, too, sounds as if agents are people. But the same sentence then says �each object is considered to be an agent...� So, are objects agents or agents objects? And if an experiment is producing a disturbance, whose perception is being disturbed -- the object�s, the agent�s, the experimenter�s, or that of some still unnamed person?
In Figure 1, the agent is shown as a circle. Is a circle an object? And it says that �the velocity of the objects is shown...� I do not see any indication of velocity in the figure. And if it were shown, in what way would it be an aid to users? And how do users fit into the cast of characters introduced so far -- agents, objects, circles?
There are places (for example, the last paragraph in the left column of page 2) where the authors seem to be describing interpersonal communication: �The agents produce a continuous probability� that someone will select one of them. But other sentences seem to me to weaken that possibility. Any guesses?
Page 3, left. What is the name (for oral use) of the little rectangle standing on its short side? I will call it �rect,� but I will also read it as variance.
In Line 2 of the left column, for what does rect-sub-si stand? The �s� first appeared on page 2 left bottom, where I took it to stand for �selecting� as in the probability p in selecting (s) a particular agent i. But now on page 3, we have rect-sub-si. I guess the subscript si means the position of an object (see bottom of the right column on page 2) with the �white noise� added. Am I correct? That interpretation fits Marken�s articles on intention. In fact, W&M-S reproduce Marken�s ratio of variances as a measure of control.
And next, we have on page 3 at upper left the rect appearing three times with no subscript! That confuses me, particularly in the sentence beginning �Tuning.� Can somebody tell me what is missing there? And the sentence after that, too, confuses me. I think of a probability as a quantity you discover (observe), not something you assign.
Can anybody tell me the meaning of �high-frequency spectral content�? I know a little about frequencies of electromagnetic waves, and I know the term �spectrum,� but �high-frequency spectral content� is beyond me.
Now we get to Figures 3, 4, and 5. The caption under Fig. 3 mentions �correct selection events,� and the text in the left column just to the left of Fig. 3�s caption says that a selection event is tallied when the probability of selecting something exceeds a criterion. What does �correctness� have to do with that?
Are the peaks in the figures occasions of selection events? Are they produced by the same person (user) or different persons? Is some one of the peaks produced by one person or several? What is signified by the various colors of the traces? What is signified by the sail-like shapes of the peaks?
The caption contains a formula for entropy. The formula contains the term x-sub-i. What does x stand for? I can find mention of it no place in the text.
In my copy, in Fig. 3, rects appear over-printed on the words �with� and �correct.� Do those rects belong someplace in that caption, or are they irrelevant?
When mentioning control, the authors in no place state control OF WHAT. I do not demand that they explain control theory. I do think some readers other than PCTers might think the omission curious. Do you?
How can I cite this paper in what I write? Judging from a few words at the bottom left on the first page, this paper was read at a convention. (I am glad I could read it on paper instead of having to try to make sense of what I could get through my ears!) But I do not know for what �CHI� stands. I suppose one could cite it through the internet. What would be the proper format?
I will be grateful for any help anyone wishes to offer. Thanks.
--Phil R
[Martin Taylor 2005.05.27.23.05]
This is Phil Runkel replying to Rick Marken's (22 May) enthusiasm about the paper by Williamson and Murray-Smith mentioned by Martin Taylor.
This communication is addressed primarily to Rick Marken. I hope he can answer some of my questions about the article. But I doubt he can answer all of them, and perhaps there are others reading this who can answer some that Rick cannot.
I'll try one or two of them.
Page 2, left column, line 7. What is an "agent"? "Agents try to ..." sounds as if agents are people. And the last paragraph in that column says, "Viewed from an agent perspective..." That, too, sounds as if agents are people. But the same sentence then says "each object is considered to be an agent..." So, are objects agents or agents objects?
I've found it useful in considering human-computer interaction to deal with active elemtns within the computer (agents) as if they had some of the communicative attributes of humans. Layered Protocol Theory came originally from that, and only later was extended to human-human communication. So I'm quite happy with the idea that the "objects" seen on the screen as being the visible manifestation of an "agent" in the computer, in the same way that a person's body is the visible manifestation of the person.
And if an experiment is producing a disturbance, whose perception is being disturbed -- the object's, the agent's, the experimenter's, or that of some still unnamed person?
I don't think they are using the word in a PCT sense -- or at least they are, but only tangentially. "The objects are subjected to a disturbance" translates to "the objects are moved randomly so that their movement would disturb the user's perception if the position of thatobject were being controlled by the user", but that's pretty long-winded.
In Figure 1, the agent is shown as a circle. Is a circle an object? And it says that "the velocity of the objects is shown..." I do not see any indication of velocity in the figure.
Yes, the circle is an object. I think the purple marks in the circles may be velocity indicators, though at that scale they are hard to see. Probably they are clear on the experimental screen.
And if it were shown, in what way would it be an aid to users? And how do users fit into the cast of characters introduced so far -- agents, objects, circles?
Agent move the objects, which are the circles, as a person moves her arm, which is an object. A user sees the object, and may be controlling the resulting perception.
There are places (for example, the last paragraph in the left column of page 2) where the authors seem to be describing interpersonal communication: "The agents produce a continuous probability" that someone will select one of them. But other sentences seem to me to weaken that possibility. Any guesses?
That whole paragraph is very badly worded, but its intended meaning seems to be that a program (another agent) is taking input from both the "disturbing" influences on all the objects and from the actual movements of the objects, and from those data is continuously computing a probability for each object that the user is controlling the position of that object.
Page 3, left. What is the name (for oral use) of the little rectangle standing on its short side? I will call it "rect," but I will also read it as variance.
It's a lower-case sigma on my screen!
In Line 2 of the left column, for what does rect-sub-si stand? The "s" first appeared on page 2 left bottom, where I took it to stand for "selecting" as in the probability p in selecting (s) a particular agent i. But now on page 3, we have rect-sub-si. I guess the subscript si means the position of an object (see bottom of the right column on page 2) with the "white noise" added.
I think it includes also the result of the user's control actions. sigme-sub-fi has just the white noise. Again, the writing is bad.
Am I correct? That interpretation fits Marken's articles on intention. In fact, W&M-S reproduce Marken's ratio of variances as a measure of control.
And next, we have on page 3 at upper left the rect appearing three times with no subscript!
Aha! The "rect" appears for all the Greek letters, not just sigma! No wonder you are confused. I think you would have no problem if you could install the appropriate "symbol" font, or perhaps a Greek one. Probably it's on your computer, but for some reason not being used to display the PDF.
That confuses me, particularly in the sentence beginning "Tuning." Can somebody tell me what is missing there? And the sentence after that, too, confuses me. I think of a probability as a quantity you discover (observe), not something you assign.
It's also something you can compute, and the computational process does the assignment. (Parenthetically, I dispute the notion that you can ever "observe" a probability).
Can anybody tell me the meaning of "high-frequency spectral content"? I know a little about frequencies of electromagnetic waves, and I know the term "spectrum," but "high-frequency spectral content" is beyond me.
It's the same as with electromangetic waves. What they mean is that the object in question may be moving slowly, but isn't jiggling fast, as the others would be, because the human is countering the fast jiggles (I guess it's not _too_ fast!)
Now we get to Figures 3, 4, and 5. The caption under Fig. 3 mentions "correct selection events," and the text in the left column just to the left of Fig. 3's caption says that a selection event is tallied when the probability of selecting something exceeds a criterion. What does "correctness' have to do with that?
If I understand properly, a "selection event" occurs when the threshold is reached. A "correct selection event" occurs when the one selected is the one the user intended. But that's reading a little between the lines.
Are the peaks in the figures occasions of selection events? Are they produced by the same person (user) or different persons?
It looks like one 18-second example trace from a single run, to show what such traces look like.
What is signified by the various colors of the traces?
I imagine each colour to represent the assigned probabiity for one of the ten objects.
What is signified by the sail-like shapes of the peaks?
That's the way this algorithm works. When the user isn't controlling any of the objects, they all have approximately the same probability (the commin red line). As soon as the user does start to control, the probability for that one starts going ujp (and also for some of the others, apparently. There can't be very many degrees of freedom for the "disturbance" or the different objects wouldn't have such correlated movements.
The caption contains a formula for entropy. The formula contains the term x-sub-i. What does x stand for? I can find mention of it no place in the text.
X-sub-i refers to the i'th object. It's sloppy, but I imagine they wanted to use the standard general formula, rather than introducing their own object-naming convention, since if the formula means anything to a reader, they've probably seen it in that form.
In my copy, in Fig. 3, rects appear over-printed on the words "with" and "correct." Do those rects belong someplace in that caption, or are they irrelevant?
Probably the characters for the big opening and closing brackets.
When mentioning control, the authors in no place state control OF WHAT. I do not demand that they explain control theory. I do think some readers other than PCTers might think the omission curious. Do you?
Not really. If you know PCT, there's no problem (and they cite Powers). If you don't, a side discussion about the fact you control perceptions would be a confusing distraction. They want to show a technique that works, not the theory behind it.
How can I cite this paper in what I write? Judging from a few words at the bottom left on the first page, this paper was read at a convention. (I am glad I could read it on paper instead of having to try to make sense of what I could get through my ears!) But I do not know for what "CHI" stands.
"Combuter-Human Interaction" The conference is called CHI 2004.
I suppose one could cite it through the internet. What would be the proper format?
I'm not good with formal formats, but I suspect that something like "in Proceedings of CHI 2004, Vienna Austria, ACM publication 1-58113-703-6/04/0004" would not be too far wrong. I won't swear to it, but I think it's likely to be refereed, and thus part of "the literature."
I hope some of this helps a bit, and hasn't made it even more confusing.
Martin
This is Phil Runkel at 10 pm on 27 May replying to Martin Taylor's 2005.05.27.23.05 reply to me.
Thanks very much, Martin. Your comments do help. I will study them tomorrow. I can say now, however, that I think your comments strengthen my confidence that the authors do know what they were up to.
Thanks very much. --Phil R
[From Rick Marken (2005.05.28.1440)]
Martin Taylor (2005.05.27.23.05)
This is Phil Runkel replying to Rick Marken's (22 May) enthusiasm
about the paper by Williamson and Murray-Smith mentioned by Martin
Taylor.This communication is addressed primarily to Rick Marken.
I'll try one or two of them.
And if an experiment is producing a disturbance, whose perception
is being disturbed -- the object's, the agent's, the experimenter's,
or that of some still unnamed person?
I don't think they are using the word in a PCT sense
Actually, they are using the word "disturbance" just as we do in PCT.
They
are applying disturbances to the positions of screen objects just as I do in
my mind reading demo at
http://www.mindreadings.com/ControlDemo/ThreeTrack.html
There are places (for example, the last paragraph in the left column
of page 2) where the authors seem to be describing interpersonal
communication: "The agents produce a continuous probability"
that
someone will select one of them. But other sentences seem to me to
weaken that possibility. Any guesses?
I believe what they are saying is that, based on their measure of stability
for each object (which, in their case, is the ratio of observed to expected
variance of object position) they compute a probability that the object is
under control. As in my mind reading demo, they continuously look to see
if the "probability of control" measure for each object exceeds a threshold.
When it does, their decision algorithm (which is a simple version of "the
test") concludes that that object is under control.
Now we get to Figures 3, 4, and 5. The caption under Fig. 3
mentions "correct selection events," and the text in the left
column
just to the left of Fig. 3's caption says that a selection event is
tallied when the probability of selecting something exceeds a
criterion. What does "correctness' have to do with that?
If I understand properly, a "selection event" occurs when the
threshold is reached. A "correct selection event" occurs when the
one
selected is the one the user intended. But that's reading a little
between the lines.
I think your reading on this is correct, Martin.
What is signified by the sail-like shapes of the peaks?
When mentioning control, the authors in no place state control OF
WHAT. I do not demand that they explain control theory. I do think
some readers other than PCTers might think the omission curious. Do
you?
Not really.
I think it is an important omission because there may be other variables
(besides the one they are using, which is position on the screen, the one I
use in the mind reading demo), such as rate of change in movement or
movement relative to other objects, that may give them better results.
I'm in touch with the authors now and I'll discuss this with them.
Best regards
Rick
Rick Marken
MindReadings.com
310 729-1400
This is Phil Runkel replying to Rick Marken's of 2005.05.28.1440:
Thanks very much for your comments on my puzzlements over W&M-S and on M Taylor's comments. I have one query left. I'll send it to the two of you directly, not via CSGnet. --P