Profile

Cover photo
Alex Holcombe
Works at University of Sydney
Attended Harvard University
781 followers|43,600 views
AboutPostsPhotosYouTube

Stream

Alex Holcombe

Shared publicly  - 
1
Add a comment...

Alex Holcombe

Shared publicly  - 
 
"Vision Research (ELS) Request to Review:
Hi Alex,
I know this is a long shot, given your dedication to open access journals. Your article (linked from your web page) is really on point, and the times they are a changing. If I didn't have such a life long fondness and respect for Vision Research I would quit the editorial board. However, if you would be willing to give free labor once more to this hugely profitable cartel, I would appreciate it. Or suggestions if you have any.
cheers,
[a Vision Research editor]"
3
Mike Taylor's profile photoJP de Ruiter's profile photoAlex Holcombe's profile photo
10 comments
 
Yes, it certainly is ridiculous. See my very first mainstream-media article on open-access issues, Peers, review your actions at http://www.timeshighereducation.co.uk/417576.article
Add a comment...
 
Edge 2014 question: What scientific idea is ready for retirement? Me: idea that Science is Self-Correcting. http://t.co/gfJva3pg75 You can't comment on the #EdgeQ   site, but I'd love to hear comments (e.g., here) 
6
1
Matt McIrvin's profile photoOpen Science's profile photo
 
Good essay, though I suppose actually retiring the idea of self-correcting science would mean giving up on science entirely, not doing things to fix the situation.
Add a comment...

Alex Holcombe

Shared publicly  - 
 
Great on relation among probability, odds, log odds, and Bayesian http://lesswrong.com/lw/mp/0_and_1_are_not_probabilities/ " Let's say that I roll a six-sided die:  If any face except 1 comes up, there's an 10% chance of hearing a bell, but if the face 1 comes up, there's a 20% chance of hearing the bell.  Now I roll the die, and hear a bell.  What are the odds that the face showing is 1?  Well, the prior odds are 1:5 (corresponding to the real number 1/5 = 0.20) and the likelihood ratio is 0.2:0.1 (corresponding to the real number 2) and I can just multiply these two together to get the posterior odds 2:5 (corresponding to the real number 2/5 or 0.40).  "
Followup to:  Infinite Certainty  1, 2, and 3 are all integers, and so is -4.  If you keep counting up, or keep counting down, you're bound to encounter a whole lot more integers.  You will not, however, encounter anything ca...
1
Add a comment...

Alex Holcombe

Shared publicly  - 
 
More news about Elsevier, copyright, and takedown notices:

"The University of Calgary has just sent this notice to all staff:
The University of Calgary has been contacted by a company representing the publisher, Elsevier Reed, regarding certain Elsevier journal articles posted on our publicly accessible university web pages. We have been provided with examples of these articles and reviewed the situation. Elsevier has put the University of Calgary on notice that these publicly posted Elsevier journal articles are an infringement of Elsevier Reed’s copyright and must be taken down."
1
Add a comment...
In his circles
310 people
Have him in circles
781 people
John Clevenger's profile photo

Alex Holcombe

Shared publicly  - 
 
Our university is trialling an Electronic Lab Notebook software, LabArchives. Someone asked me why I won't be using it, FYI here's my answer.

I'm more inclined to support open-source solutions. But open-source type solutions are currently harder to use I think, so for those who don't have the technical know-how and don't have time to acquire such skills, LabArchives may be the way to go.

For myself, I favor Github and also Open Science Framework for code and data archiving, and also for code management in the case of github.  For example, for the project I'm currently working on, I'm in a data analysis phase and am using github for all the version control through R, here: https://github.com/alexholcombe/speed-tf-VSS14 .

For archiving data, I'm currently inclined to use Open Science Framework, e.g. program code and data for experiments associated with an in-revision paper is here : https://osf.io/t4vmy/
OSF actually doesn't require technical know-how at all (although they're adding integration with github, which I may use in future), but it's mainly just a place to drop files currently.

Neither GitHub nor OSF have daily-lab-notebook type functionality, but I don't need that because all the parameters and details associated with running a subject in an experiment are automatically saved by my PsychoPy program when I execute it, including a frozen copy of the code at the time it was run and system info.  For those using equipment not integrated together, like say a neuropsychological test done with paper and pencil, an EEG run somewhere else, and a behavioral test, a lab notebook might make more sense.
4
4
Darren Finlay's profile photoMatthew Todd's profile photoGuy Bilodeau's profile photoHaroon Rasheed's profile photo
 
As you know we use LabTrove, an open source alternative, but I think sometimes large organisations prefer software as a solution as opposed to something with the promise/threat of long-term investment in, for example, actually developing software. It's interesting that the lab notebook in Chemistry is traditionally like a diary - the entries are in date order and I think subconsciously we sort the experiments along a timeline ("I remember I did that experiment back in December"), but with a move to electronic the book is finally searchable - joy. I guess I'm surprised you don't do the same for e.g. testing flashy colours on people, i.e. "Today subject X was unable to distinguish colours at Y rpm, unlike last week when Z happened."
Add a comment...
 
Hi +Open Source Malaria , do you have an explicit license for your data beyond saying CC-BY?  

The Open Knowledge Foundation has created the Open Data Commons Attribution License (ODC-BY, http://opendatacommons.org/licenses/by/summary/), and it is roughly what we want for data associated with the Registered Replication Reports (RRR) article type of Perspectives on Psychological Science.  

I am not clear on whether there is any difference between conventional CC-BY and ODC-BY, perhaps there's not.

For RRRs, the data collected by each lab is posted on their respective OSF pages.  Our policy is still in discussion but may end up being that other researchers can use the data freely, but only after the RRR has been published-  an embargo on use of the data. This is not an option with ODC-BY, although perhaps an informal add-on clause would be sufficient.  
1
Matthew Todd's profile photo
 
Don't know. Appears similar - is there a need for two things?
Add a comment...

Alex Holcombe

Shared publicly  - 
 
"Sperling, 1963. Human Factors, 5, 19-31 --one of the five most cited articles in the history of the journal). An interesting sidelight is that this research was carried out in the Psychology Department after I had been failed by Harvard's Social Relations Department and was not permitted to carry out the project I had originally planned." http://www.opt.uh.edu/files/opticalSociety/2013_OSA_VM_Program.pdf
2
Add a comment...

Alex Holcombe

Shared publicly  - 
 
Thomas Huxley wrote to his sister in 1852,

“You have no idea of the intrigues that
go on in this blessed world of science.
Science is, I fear, no purer than any other region
of human activity; though it should be … So I
must manoeuvre a little to get my poor memoir
kept out of his [a competing scientist’s] hands”

(Huxley, 1852). Via James C. Coyne
3
1
Jon Brock's profile photo
Add a comment...

Alex Holcombe

Shared publicly  - 
 
What is the probability of seeing something you have never seen before?

This is interesting for people who study perception, because a currently-popular theoretical approach to object recognition etc. is Bayesian modeling, under which one must assign a prior probability to all possible objects. The way models are usually set up, no allowance is made or probability assigned to completely novel objects, so they will always be misclassified as an already-known object. You need to know when something is completely new so that you can create a new item in your object lexicon.

Laplace and later Alan Turing came up with ways to estimate the probability of encountering an unknown object.  As described by Orlitsky et al., "Good and Turing encountered this problem while trying to break the Enigma cipher during the Second World War. British intelligence was in possession of the Kenngruppenbuch, the German cipher book that contained all possible secret keys, and used previously decrypted messages to document the page numbers of keys used by various U-boat commanders. They wanted to use this information to estimate the distributions of pages that each U-boat commander picked secret keys from." In other words, what is the probability of each page number being used, even page numbers that had never been seen to be used before?

What Turing came up with is explained by Weng & Owens & Lee:
"The total probability mass assigned to unseen events is the
same as the relative occurrence of words [objects] that appear just once! This makes sense, because appearing zero times is not so different from appearing once in a relatively small sample". Turns  out this estimator can be derived in multiple ways:  http://www.cs.cornell.edu/courses/cs6740/2010sp/guides/lec11.pdf 

The details are pretty cool: http://www.sciencemag.org/content/302/5644/427.full Thanks to  David Wipf for putting me on to this Good-Turing estimator. I should add that even with this help from Good & Turing, I'm not convinced that the brain is doing Bayesian estimation. I think perception is a complex dynamical system that can throw up all sorts of percepts, with the constraints of prior experience being weaker than a Bayesian approach involves.
3
2
Vasileios Tatsis's profile photoPetros G. Sideris's profile photoMatthew Baggott's profile photo
 
You win my personal award for the best timing of the day. I have been using a Laplacian estimator, but am not happy with the results I got today and decided to read up on it more. These references are just what I need! 
Add a comment...
People
In his circles
310 people
Have him in circles
781 people
John Clevenger's profile photo
Work
Employment
  • University of Sydney
    Associate Professor, present
Basic Information
Gender
Male
Story
Tagline
Academic. Doing basic research, and improving the system of science.
Introduction
Education
  • Harvard University
    Psychology, 1995 - 2000
  • University of Virginia
    Cognitive Science, Psychology, 1991 - 1995