Perceptual Inference Predicts Contextual Modulations of Sensory Responses

Another predictive coding paper (by Udo Ernst and others)

 

 

Aan u verzonden door Sander via Google Reader:

 

 

Perceptual Inference Predicts Contextual Modulations of Sensory Responses

via Journal of Neuroscience current issue door Lochmann, T., Ernst, U. A., Deneve, S. op 21-3-12

Sensory receptive fields (RFs) vary as a function of stimulus properties and measurement methods. Previous stimuli or surrounding stimuli facilitate, suppress, or change the selectivity of sensory neurons' responses. Here, we propose that these spatiotemporal contextual dependencies are signatures of efficient perceptual inference and can be explained by a single neural mechanism, input targeted divisive inhibition. To respond both selectively and reliably, sensory neurons should behave as active predictors rather than passive filters. In particular, they should remove input they can predict ("explain away") from the synaptic inputs to all other neurons. This implies that RFs are constantly and dynamically reshaped by the spatial and temporal context, while the true selectivity of sensory neurons resides in their "predictive field." This approach motivates a reinvestigation of sensory representations and particularly the role and specificity of surround suppression and adaptation in sensory areas.

 

 

Dingen die u vanaf hier kunt doen:

• Abonneren op Journal of Neuroscience current issue met Google Reader
• Aan de slag met Google Reader om eenvoudig al uw favoriete sites bij te houden

 

 

Computer Vision online course

While some of you may be following Udacity courses on Python (I hear excellent comments about those), Berkeley is gonna offer soon ("Class starts March 2012") a Computer Vision course taught my Jitendra Malik whose work is both fantastic and relevant for us. I signed up!

There are even more high-quality courses offered, check the following:

- Udacity

- Coursera

- MITx

The difference between freely accessible content and Open Access

 

 

Aan u verzonden door Sander via Google Reader:

 

 

The difference between freely accessible content and Open Access

via cognitive science and more door Sebastiaan Mathôt op 12-3-12

'Open access' is somewhat of a buzzword within academia. But unfortunately the term is widely misunderstood. All too often, the label 'open access' is used to refer to any content that can be accessed online for free. By this (incorrect) definition, most websites would be open access, because they offer content that you can read (or watch, listen to, etc.) for free. But you are usually not allowed to reproduce that content, or to sell it, modify it, etc. The only thing that you, as a consumer, are allowed to do is view the website right then and there, for as long as the owner of the website permits you.

This may be free of charge. It may even be a nice gesture. But it is not open access.

The definition of 'open access' is still evolving, but by consensus a number of conditions need to be satisfied for it to apply. In PLoS Biology, Michael Carroll gives us the following list:

Full open access content is

• Easily accessible online
• Available to anyone free of charge
• Available for re-use without restriction except that attribution be given to the source.

No one of these alone qualifies content for an open access label.

The paper by Carroll satisfies these criteria. I downloaded it for free. I am allowed to reproduce and redistribute the paper in it's entirety (here it is). And so are you. I didn't need Carroll's permission to upload his paper. And I do not (have to) care if he likes it. I just needed to provide attribution.

An open access license (such as CC BY) provides safeguards. Once a scientific paper is published under an open access license, it will be available to anyone, for ever (unless, of course, all copies get lost by some bizarre coincidence). Neither the author nor the publisher can undo this afterwards: The work is safe. In addition, open access publishing makes science accessible to the general public, which is invaluable. (Scientists are, after all, public servants.)

 

 

Dingen die u vanaf hier kunt doen:

• Abonneren op cognitive science and more met Google Reader
• Aan de slag met Google Reader om eenvoudig al uw favoriete sites bij te houden

 

 

open access at KU Leuven

Unfortunately, the new initiative for open access does not seem to
mean that the KU Leuven is now offering funds to publish papers in
open access journals.

http://www.kuleuven.be/onderzoek/open_access/index.html

Sign the White House’s public access petition!

 

 

Aan u verzonden door Sander via Google Reader:

 

 

Sign the White House’s public access petition!

via Sauropod Vertebra Picture of the Week #AcademicSpring door Mike Taylor op 29-2-12

A quick note to remind everyone that although the RWA is dead, that only brings us back to the status quo.  At present, it's still the case that the great majority of US government-funded research goes behind paywalls.  Although the National Institutes of Health (NIH) has a public access policy that is resulting in a lot of papers being posted for general access at PubMed Central, the NIH is only one of a dozen U.S. Federal Agencies with research budgets exceeding $100 million.  The others are:

• Department of Agriculture
• Department of Commerce
• Department of Defense
• Department of Education
• Department of Energy
• Department of Health and Human Services
• Department of Homeland Security
• Department of Transportation
• Environmental Protection Agency
• National Aeronautics and Space Administration
• National Science Foundation

Wouldn't it be great if all those agencies had similar policies?  If all the research funded by any of those agencies had to be openly accessible not only to all researchers but to the public — teachers, nurses, artists, translators.

That is exactly what the Federal Research Public Access Act (FRPAA) will do if it passes.  In the RWA backlash, we have a unique opportunity to rally support and ensure that this important bill passes, despite the handicap of having been proposed during an election year.

What can you do to help?  First, sign the petition at whitehouse.gov.  I've signed it: turns out you don't need to be a U.S. citizen for your voice to be heard.  It takes a minute to register on the site, then a second to sign.  Stop reading this post and do it now.

Second, if you are a U.S. Citizen, you can contact your representatives to express your support and solicit theirs.  For more on this, see the Alliance for Taxpayer Access's page.

And whoever you are, you can spread the word.  Blog.  Tweet.  If you're at a university, raise the subject with your colleagues.  If you're on job-search or tenure committees, undercut barrier-based publishing's historic advantage by rewarding candidates for the quality of their work rather than the journal it's published in.  (One simple way to do this, though far from perfect, is to look at citation counts rather than impact factors.)

As Michael Eisen has said, we won the Battle of the Research Works Act. Now let's win the War for Open Access.

 

 

Dingen die u vanaf hier kunt doen:

• Abonneren op Sauropod Vertebra Picture of the Week #AcademicSpring met Google Reader
• Aan de slag met Google Reader om eenvoudig al uw favoriete sites bij te houden

 

 

Workshop on Simplicity and Likelihood in Perceptual Organization

Understanding intrinsic simplicity and statistical likelihood in visual Gestalts

March 14-16, 2012, University of Leuven, Belgium

The Laboratory of Experimental Psychology is hosting a small-scale
workshop on simplicity and likelihood principles in perceptual
organization. The workshop aims at understanding how the brain
organizes the visual input into a coherent interpretation. Why are
certain organizations preferred over others? Is it because those
organizations are simpler, or because they are more likely?

The workshop combines tutorial style lectures by our invited speakers
(Jacob Feldman, Lars Muckli, Peter van der Helm, & Laurenz Wiskott),
with short empirically focused presentations by lab members (Hans Op
de Beeck, Wouter Braet, Maarten Demeyer, Lee de-Wit, Jonas Kubilius,
Bart Machilsen, Tom Putzeys, Michaël Sassi, Sander van de Cruys,
Nathalie van Humbeeck, Cees van Leeuwen, & Johan Wagemans). There will
be ample time for open discussion.

Anyone who is interested in participating is welcome, but note that
attendance will be limited to 30 places. Registration cost is 50
Euros. More details and practical information about the event can be
found here:

http://www.gestaltrevision.be/en/events/upcoming/27-events/234

AVA/BMVA spring (AGM) meeting 2012

Microsoft Research, Cambridge
May 22, 2012 – May 22, 2012

AVA/BMVA Meeting on Biological and Computer Vision

Microsoft Research, Cambridge

Abstract submission closes: 30th March 2012

Invited Keynote Speakers:

Aude Oliva (MIT)

The study of biological and machine vision share much common history
(eg Marr), and each discipline has benefited enormously from findings
and techniques from the other. The aim of this meeting, organised
jointly by the AVA (UK biological vision) and BMVA (UK computer
vision) is to reignite conversations between these two fields.

We particularly encourage submissions that will be of potential
cross-disciplinary interest to both human and computer vision.
However, regular human vision AVA submissions, as well as computer
vision techniques likely to be of interest to a general audience, are
welcome.

250 word short abstracts can be submitted via the conference website.
For the most part abstracts will be reviewed and published in the
journal Perception and should therefore describe original work.
Authors may choose to present less original cross-over work or to
withhold their abstract from publication in which case they should
choose the "no publication" option when submitting.

You can submit an abstract from the "call" page.

The registration fees depends on whether you are a member of the AVA,
the BMVA or neither. See the registration link. Note that AVA and BMVA
have different funding models, hence the different prices.

http://www.theava.net/conf/index.php?conference=Meeting&schedConf=avabmva