FOR GREAT SCIENCE!

I got the following email today. If you recall, a year ago I put up a scientific survey for an actual scientist friend of mine. (I’ve toured his lab, he is doing some crazy Matrix brain stuff in there).

He is doing some fantastic work though, which has the potential to help millions of people. So if you’d like to help out, the link is below.

***

Hi folks:

It seems our occasional cabal of writers has taken a hiatus from rabble-rousing.  So I would like to take this opportunity to thank everyone who promoted my lab survey last year.  We received over 1100 individual responses to the survey, which kicked the results well into the “statistically significant” territory.  [1100 responses / 20 different surveys meant 50-60 responses per survey image – the statistical goal was >30 per image.]

Over the last year we’ve been working with that data – and it is *great* – but we discovered that a database of just 500 pictures is not enough.  We need 1000.  So I’ve put together a new survey of another 500 pictures – again, there’s only 25 pictures per survey, but if enough people take the survey we will get significant results.

I’d love to have your help again this year to promote the survey.  Larry’s MHI fans alone probably promoted around half of the responses.

The info to announce the survey is below – or you can link/share my recent post on Facebook.

Thanks all!
-Speaker / s2la, Speaker to Lab Animals
_____________________________

Dear Friends and Colleagues:

Last year, the Hampson Laboratory Webpage (http://hampsonlab.org), conducted a survey of 500 pictures.  We received over 1100 individual responses, for which we are extremely grateful to all who participated, and the people who rallied their own social media groups to our benefit.

This year we have another 500 pictures to classify.   We are asking for volunteers to go to the Hampsonlab.org page and take a survey consisting of 25 images from our set.

Our laboratory is identifying “categories” and/or “features” of pictures that we use to examine how the brain encodes information according to a number of different characteristics – is it a cartoon? Photograph?  Silhouette or drawing?  Is it in color or black & white – if color, which colors?  Are there specific items visible in the picture?

We know that different people categorize pictures in different ways.  Thus, we need to conduct a survey of as many people as possible to find the most likely common classification from a fixed set of categories.

There are instructions and hints on the URL web page (http://hampsonlab.org). Clicking on “Take the Survey” will bring up a random selection of 25 pictures.  Enter your responses by clicking next to the features that you think fit the image.  The listed features will not perfectly suit all pictures.  We know this.  The features were chosen for reasons based on the psychology and physiology of human memory.  Therefore, we ask that you choose the closest match(es) from the list of options.  At the end of the page, “Submit” the Survey, and your responses will be written to our server.
If you have time, click on “Take Another Survey” and the webpage will return to the beginning.  Each time you click “Take the Survey,” you should see a new page of 25 pictures. [You can choose your survey page by bypassing the default screen and editing the URL to read: http://hampsonlab.org/v3Survey1.html , …v3Survey2.html, …v3Survey3.html, etc. through  …v3Survey20.html.  Again, all results are logged on our server automatically once you select “Submit.”]

Disclaimer:  The survey is completely anonymous – we record only the picture name and 1’s or 0’s representing your choices (you can briefly see the data in the box on the Submit page).   All pictures in the survey are purchased or used under fair use, non-commercial research purposes only.  Your response data contains no personal information.  We conduct no diagnosis or analysis of the participants or individual responses.  The data is used solely to develop anonymous population classifications.

Thank you for participating in the survey.  We appreciate your help.

Peter Grant wins the 2nd Annual CLFA Award
Charity Gaming Event on March 30th in Draper, Utah, with Me and Tracy Hickman

19 thoughts on “FOR GREAT SCIENCE!”

    1. I *could* share a matrix of results – but then it becomes “research” and I have to jump through a number of regulatory hoops for approval. Right now, it is simply an anonymous survey with no feedback, diagnosis or benefit to the participants , thus exempt by human research oversight rules.

      To add to the problem, it’s a huge spreadsheet – with over 40,000 lines, condensed to a 1000-line pivot table.

      What I *can* tell you is that what is important to us is finding categories of images that can be distinguished as: >80% of people think this image is Category A, <20% of people think this second image is Category A, around 50% of people think this third image is Category A.

      This is why categories and colors are limited. We chose features and categories that we know from other research are "high level" categories specifically encoded by the section of the brain we study. We know that not every image will fit those categories, but it's important to us to be able to identify which pictures are definitely *in*, definitely *out* and *possibly* in a category.

      1. Cool. I’d be interested in it from more of the machine learning and image recognition side of things. Unfortunately your survey results are on the small side for machine training data, but I’d be interested to see how an image recognition neural net compares to the human results you gathered.

      2. You! You’re one of the survey creators! Here’s a question you can answer without screwing the pool, bureaucracy-wise.

        Why no “brown” as a colour choice? Is this secretly a test to make artists / the literal-minded crazy?

        I’m not joking. I can take one go-round of your survey for science’s sake, and the Elephant’s Child, but then I’m done.

    1. Unfortunately, I got hacked several years ago. I altered security and cleaned up the site, but I guess that hasn’t filtered through all of the security. If you tell me which software is reporting my site, I can petition them to release it. Kacey Ezell tells me that DoD networks also block access, but that may be for other reasons (i.e. the .ORG address).

  1. I ran into an issue when doing the survey. That issue was at times, none of the colors listed as choices were in the picture.

    I asked s2la about it on the Baen forum. He said if no proper selection available in a category just leave it blank. That won’t invalidate your efforts. It’s all about how you, the individual, process and organize data.

    1. yep, the most frequent color I wanted to enter was ‘brown’ followed by ‘white’. I left them blank, but it was a bit frustrating. I SOOOO wanted to pick a color!

      1. I know, right?

        Knowing s2la a bit from the Baen Bar forum, I kinda sorta suspect that the real purpose of the study is charting the time delay as we try to figure out how to get that damn color pick done.

  2. Hey Speaker, one comment about the survey. On many of the pictures, if I clicked “Open picture in new tab”, the natural size of the picture was larger than the image box presented — and I could sometimes see things in the larger sized picture than in the smaller-sized version presented on the page. For example, a few pictures that looked like landscapes with no people present in the smaller size, when I looked at them in the larger size, had people or vehicles just barely visible in the distance.

    I chose to answer based on the smaller-sized images presented on the survey page, since I assumed that that’s what would give you the best results. But since there was no specific instruction saying “You might be able to zoom in and see more details, but please don’t: answer based only on what you can see in the image as-presented”, I wanted to mention it. Depending on how you use these images later, I could see that being an important factor to consider: whether or not certain details are visible at the specific zoom level being presented.

  3. Oh. It’s these guys again. The rat bastards who won’t include “Brown” on a colour palette.

    I want to stab them in the face, Patrick.

  4. Just finished the survey, though I may go back and do another set. One thing that occurred to me while I was taking it was that people trained in photography or other visual arts may classify an image differently because of their training / experience. Since you don’t collect any kind of personal data (kudos) I assume you’re hoping to mitigate the effects of this or similar biases with a large enough sample size. Good luck, and glad I could help.

Leave a Reply to Old NFO Cancel reply

Your email address will not be published. Required fields are marked *