DSPRelated.com
Forums

Subliminal messages real or not? (OT)

Started by NightHawk July 24, 2005

Leon wrote:
> My degree is in experimental psychology (and maths). One of our > lecturers did a lot of research into "subliminal perception", > attempting to replicate many of the published experiments, and was > unable to obtain any statistically significant results. He had a lot of > problems getting his papers published.
What was the reason for not getting published? That he dealt with subliminal messages, or that he could find no effects from them? Rune

Dave Coffey wrote:
> I agree with Rune that it may be nigh on impossible to > detect by strict DSP means (assuming of course that such a > phenomenon exists). > > The human auditory system is however, a different kettle > of fish. A number of years ago, I listened to a very > interesting interview on radio, where the interviewee > was involved in researching intelligibility limits in > human hearing. Anyway, the crux of the interview was > a demonstration, where a short passage (1-2 secs) of > noise was played back. The interviewee then uterred a > sentence and played the noise passage again. On the second > pass, I could clearly hear the word 'Apple' in the noise. > He had in some context used the word apple in the sentence > he spoke in between the two passages. > > No specific details of the research or the suspected mechanisms > were given in the interview (and I'd had x+1 Barfo's that evening, > - it was Summer :-) ). My own (humble) take on what was going on, > was that the neural pathway involved in processing the spoken > word 'Apple' had remained sensitised for a while and subsequently > acted like a matched filter on the second noise passage so as to > significantly increase the (perceived) SNR of the passage. > > I'm no perceptual/psychoacoustic modelling expert, some people here > on this forum are much smarter than me in this fascinating area. > However, to me, this sounds more like a job for Neural Networks > with extremely fast pattern matching capabilities. > > Dave.
I've seen the same demo in the context of seismic data. Processing and analyzing seismic data is about making sense of hints of lines and "blobs" in a noisy image. Here is an actual seismic image (don't ask me what it is or how to interpret it, though): http://www.geol.vt.edu/research/3d-exxon.gif The infamous image of the dalmation, http://www.bbc.co.uk/radio4/reith2003/images/lecture3_dog1.gif is perhaps easier to see for most. Seismic processing and finding the dalmatian, is the same thing: Take a few (scarce) visual hints and make sense of them. What seismics is concerned, in the presence of noise. The problem is that the seismic sensors don't always work, and loss of data occurs, introducing arifacts in the seismic images. A few years ago, I saw a demonstration where somebody had introduced white noise where the missing data were supposed to have been. The images were clear and bright. I don't know they actually proved. I see at least two possible explanations. The first is that seismic data analysis is a matter of whishful thinking and the analyst sees what he wants, irrespective of the data he works with. I have some reservations against that view. The other possible explanation is that the redundancy in the data set is so large that the "true" image will appear if only the "holes" in the data set are treated "gently". Perhaps a little bit acceptable, from the industry's point of view. Whatever the case may be, that human brain is certainly a factor to be reckoned with when processing data. Rune
Because his results were non-significant, journals don't like
publishing such papers.

Leon

On 28 Jul 2005 21:01:24 -0700, "Leon" <leon_heller@hotmail.com> wrote:

>Because his results were non-significant, journals don't like >publishing such papers.
That's a shame. Were these real, "hard science" journals? Does this happen in other sciences? I recently read one of Richard Feynman's talks (IIRC it was from "The Pleasure of Finding Things Out"), he was saying, as part of good science, how important it is to publish negative results, so (as one example) future researchers can find out if the topic was even studied before. Nowadays one can cheaply publish on the Web, though you don't need a peer review to publish on the Web, so it won't carry as much weight as a scientific paper.
>Leon
----- http://www.mindspring.com/~benbradley
I agree. Negative results can serve a purpose.
For the reasons you mentioned and also to serve
as a basis for incremental research and discovery.

If you think publishing practices are less than
desirable in Physics/Engineering you should take
a look at the pharmaceutical industry!!

Dave.

They would have been psychology journals, like the British Journal of
Psychology or the Journal of Experimental Psychology.

Leon