Thanks, glad you like it. There's a piece of my soul in that thing. Or I assume soul-transfer is what happens when you go full hermit for 5 months and gain 50 pounds creating something like this.
Any questions about the science or AI, happy to talk about it. :)
Good for you, but that's none of my business. I don't want to be harsh, but I don't like this kind of emotional manipulation. You're trying to prime me to be buddies with you and cut you some slack.
I have some questions. In what sense is this an AI and not simply a computer program? Did you use methods that are discussed in AI or machine learning textbooks and conferences, or did you simply put the cool label AI on what is essentially "just" a hand crafted algorithm?
I know there is no clear separation as AI is a subset of algorithms, but still I feel some dishonesty going on here.
Also in other comments by the company, I can see the attitude about the science that (not precisely quoted) "we're not sure yet, we're excited to find out" and "give it a try and compare it to other music how it goes for you" etc. Well, no. You're making the claim that there is some solid science going on behind this, that you have more than a dozen studies and you put pretty, science-y charts on the website. Now I'm not saying that it doesn't work, but please show me the peer reviewed randomized double blind studies that show altered brain waves and some significant quantitative change (preferably not only self-reported things about how they "feel"). Preferably there should be different groups: no music, generic pop music, normal ambient music, some baseline of your company's music and finally the specially crafted fully optimized version of your company's music.
Until I can see that, verified also by someone who's not paid by the company, I'll have my doubts.
Someone poured their heart and soul into something that may not make a profit! Don't be such a cranky scrooge just because they confess that.
An ex used to run neurological studies, and they are incredibly expensive. The small survey groups will no doubt improve with time. If you don't believe the hard science, then try the product and see if it works for you compared to the music that's linked to in the above posts. Further criticism is unwarranted.
> Someone poured their heart and soul into something that may not make a profit!
Sorry but that's irrelevant. A homeopathy practitioner could say the same thing.
Their site front page says in big font: "Dramatically improve focus, relaxation & sleep". And it says science all over the place and contains pretty sciency charts. An intentionally wide range of charts. Vertical and horizontal bar charts, pie charts, line graphs, all to make you go "this looks legit science!"
So where is the peer-reviewed, randomized, controlled, double blind trial to support these claims? The only study that tries to compare it with placebo music seems to be unpublished as of yet and all we can see is some charts. We don't see their methodology, the results, significance etc.
I would have no problem with this if they didn't brand it so extremely heavily on science.
We're planning on following that one up with a more robust study, because our users do get great results with HRV and it's an interesting topic.
One of our former neuroscientists published a well received meta analysis in a journal, but it's a bit outdated now re: what we're doing. Still, it's on our site if you're interested.
And then of course here's our own research that is pending publication:
We also do regular analysis of user progress with a few statisticians I work with, and those will be published as well. Sorry, just takes time to get these things ready and through to a journal. Trying our best though. :)
I appreciate your honesty here. If I'm reading it right, result of the focus study is that the Music-Placebo difference is at p=0.044, 0.041, 0.048. So it is significant (under 0.05) but "just barely", which is always a warning sign, especially if the study was done and published by the company itself.
It's actually a very very good result for the tests we did. A similar study using a boatload of caffeine couldn't reach statistical significance. Also I believe strongly that if we'd compared it to say, classical music or some random ambient album, it would have been an even stronger result. The placebo in this study was still generated by the AI, so it followed the same rules - rules that naturally help people focus. The only thing it lacked was the brainwave stimulating modulations. We did see a big difference on an EEG because of that, but the fact is that part of what makes the sessions work is that they follow strict rules, and there's nothing wrong with that. It just makes it harder in this case.
But I actually like it like that. I want a double blind study with identical-sounding placebo and more subjects. After that we can study how it compares to regular music commonly used as study-aids.
Regarding the statistical analysis, result interpretation, etc, I feel I should let Giovanni himself defend his paper. He's dealing with family / health issues right now, but I hope to get him on here as soon as possible. I promise he will respond eventually and clear this up.
> I believe strongly that if we'd compared it to say, classical music or some random ambient album, it would have been an even stronger result.
Strong beliefs like that are still just hunches.
> I want a double blind study with identical-sounding placebo and more subjects. After that we can study how it compares to regular music commonly used as study-aids.
My problem is that you already decided that it "dramatically improves focus, relaxation and sleep". What if the new better study doesn't show the effect? Will you dissolve the company? Will you just change the marketing?
I'd also suggest "triple-blinding" the study, i.e. the person who does the statistical analysis on the computer should also not know which music was which.
Hey, no problem. Not trying to "prime" you here, this is just how I talk. I'm not that strategic. If you think it would help, I'd be happy to get on a call/skype? I just feel like there's some miscommunication going on here and I'd like to clear it up if I can.
The pretty charts you see are direct copies of the EEG analysis by Giovanni. We did try to make them pretty, but if you're interested in the originals check out the analyses Giovanni did. He'll be on at some point to explain them further if you want to direct some questions to him directly.
3) Your study procedure recommendation
Regarding your study procedure recommendation, that's a great idea. We did something very similar to that in the Focus analysis above. 3 conditions: No Music - Music Without Stimulation (placebo) - Music With Stimulation. To the untrained ear, the music was indistinguishable. Most people can't tell which is which, and the feedback we got from the subjects confirmed it. It was actually really interesting: some of them asked to be able to keep the placebo-music (not calling it that, obviously), saying they worked better on it, when their results often told a different story.
Next study will be double blind and have more subjects. Giovanni feels we have enough information now to do it right.
So, someone above asked about the AI thing and I'm happy to explain it. I'll paste a lot of that below. Although I have to say I'm not sure what textbooks are calling anything these days. When I talk to programmers right out of college sometimes there's a bit of confusion as they talk about patterns I know, but use different names. I grew up programming on an 8088, and I've been working for myself for 13 years, mostly by myself, so I'm sure I'm behind on a lot of things.
Having said that, I'd probably call what I'm doing "Emergent" because there are many competing little pieces. Let's see if I can explain it quickly here, though please don't be too hard on me I'm trying my best to answer a lot of questions, but I'm really happy to go more into later after all this dies down :) Conceptually, you could think of it as first creating what I'll call "song-bot", which acts a kind of overlord, and has some instructions from me (like maybe there's some specific chords I want it to use, keys to avoid, tempo, brainwave protocols, genre, etc, etc). This song-bot guy then spawns a bunch of little other bots that compete with each other for the right to play/fade in/or generally be a part of the final result. These little guys have different characteristics, like a "drum-bot" might has different places it "wants" to be placed, and so generally competes with other drums, but not always. Sometimes I'll even have little "bots" for individual notes of a certain instrument. They obey certain rules of course, to form a background, drum line, or a simple melody, and then they pass that information along to subsequent incarnations (there's some learning involved in that process, though I hesitate to give it any textbook term - man, you have me terrified here of defining something wrong! :) ). Through that learning, the pattern the original little guys made has more weight and will tend to repeat. But again, not always.
After a while, out of all these smaller, simple processes form a very complex song. Complexity "emerges." So I call it a kind of "emergent intelligence."
The resulting song can be quite complex and varied. Of course there's more to it, because songs have sections and clever variance, but all in all the genres I'm using are very structured. Techno especially, is very easy for the AI. Sometimes I'm tasked with creating a theme in a particularly difficult genre, and that's when it gets really tricky/fun (such as some of the Indian ones, which were labors of love). In these cases, I may have to re-rerun many "generations" of the AI through the same song from beginning to end, with different parameters/instruments, but with the same patterns/learning, because what can happen in these cases is that it starts out simple, and increases in complexity as it goes, so the start can sometimes be a bit boring.
Also keep in mind this is just a conceptual explanation. The code is much less amusing. Thinking of naming processes "bots" just for fun though.
I hope I explained it OK. Listen, I'm not saying it's Watson or anything, or even that I'm following the definition of "emergent intelligence." I didn't go to a textbook to make this, it's really just a necessary step I had to take in order to make the computer do what I wanted it to do. I tried out some different ideas, and this one worked the best, in my opinion at least.
> If you think it would help, I'd be happy to get on a call/skype? I just feel like there's some miscommunication going on here and I'd like to clear it up if I can.
Nah, no need. I can't say much more than what I said already. As a computer science major who took AI and machine learning courses and as a self-identifying scientific skeptic, I just felt the need to point out the fishy things going on here. Anyone here can read your reactions as well and build their opinion.
Thanks, glad you like it. There's a piece of my soul in that thing. Or I assume soul-transfer is what happens when you go full hermit for 5 months and gain 50 pounds creating something like this.
Any questions about the science or AI, happy to talk about it. :)