My pareidolia knows no bounds.

Recognizing (a picture of) the heads at Mt Rushmore as a sculpture of several human heads is NOT an error.(Recognizing it as several <u>actual</u> human heads would be an error. But that is/seems-to-be a 'lesser' error than the following.)

Recognizing (a picture of) a naked girl in the tiles on my den floor IS an error. (I just went to look - I can still see her. But that recognition is STILL an error.)

No pattern-recognizing system is, or can be, perfect. Sometimes they make mistakes.

What's the problem?

<blockquote id="quote"><font size="2" face="Verdana, Arial, Helvetica" id="quote">quote:<hr height="1" noshade id="quote"><i>Originally posted by Larry Burford</i>
<br />

Recognizing (a picture of) a naked girl in the tiles on my den floor IS an error. (I just went to look - I can still see her. But that recognition is STILL an error.)
<hr height="1" noshade id="quote"></blockquote id="quote"></font id="quote">

I don't understand why you would think that's an error. A pareidolic (ressler) image is real. Of course the naked girl isn't real, but nobody ever claimed that the objects of our pattern recognition system are real. They are just patterns that look like something to us.

If a computer PR system is looking for Bs, and finds an inkblot that it thinks is a B, that pattern fit the algorithm sufficiently for the computer to count it as a B. It's not the computer's error, it really looks like a "B". The programmers have to go back and figure out how to add some refinement to the code to make it capable of differentiating.

We never claimed there were errors involved in any of this, other than the originalists who called it a mental disorder resulting in "partial hallucinations."

Added note: I did however claim that Neil (or any AOH proponent) was in error thinking these patterns were real art. If you thought that really was a naked lady (in the flesh), well, I'd say that was an error.

rd

Using pareidolia(new) we can rule out mental disorders.

***

But a PR system can still make an erroneous recognition. You suggested a programming error. Whether the PR system in question is natural, manmade, or super natural, what else could be the cause?



But suppose for the sake of argument the programming is perfect - the error/mistaken recognition is still there. The image of B was recognized as A.

<b>[rderosa] "If you thought that really was a naked lady (in the flesh) ..."</b>

I know she isn't really there. But I still see her. My pattern recognizer still says 'naked lady'. It is malfunctioning. It is making a mistake. An error.

Isn't it?

<blockquote id="quote"><font size="2" face="Verdana, Arial, Helvetica" id="quote">quote:<hr height="1" noshade id="quote"><i>Originally posted by Larry Burford</i>
<br />
Isn't it?
<hr height="1" noshade id="quote"></blockquote id="quote"></font id="quote">I say only if you think she <b> is </b>real, but not if you just think she looks real.

But after re-reading the article that started us on this path:

www.theatlantic.com/technology/archive/2...in-computers/260760/

I do see the dilemma. I think.

In the world of computers, they are false-positives. So, does "false-positive" = "error"?

I don't think so. I think there's a difference and anyway one has to decide who is making the error, the programmer or the computer.

In the world of humans, no error is involved, since you really see the pattern that evokes the object image in your mind.

rd

<blockquote id="quote"><font size="2" face="Verdana, Arial, Helvetica" id="quote">quote:<hr height="1" noshade id="quote"><i>Originally posted by Larry Burford</i>
<br />The image of B was recognized as A.
<hr height="1" noshade id="quote"></blockquote id="quote"></font id="quote">

OK, hold the phone. I think I see what happened here. Let me try to clarify something.

Suppose a programmer wrote a Pattern Recognition algorithm that said: Find all the "A"s and "B"s in the field of view.

Then the user runs the programs on various images, some of which contain both "A"s and "B"s, or just "A"s or just "B"s.

Sometimes the PR system correctly finds "A"s, sometimes it correctly finds "B"s.

In this scenario, if the program calls an "A" a "B" or vice verse, yes that's an error.

But what if the program finds a pattern that is neither an "A" or a "B", but rather just an inkblot that looks like an "A" or a "B", that's a false-positive, or a pareidolic (ressler) "A" or "B" and <b>it's not an error.</b>

In other words, all pareidolic (ressler) images are not errors.

rd