Wednesday, October 29, 2008

Publishing in the cloud

In research the old principle of publishing or perishing has long been replaced by the more quantitative performance metric of publishing in journals with a high impact factor. For example, in the case of the last paper on which I am a coauthor (today papers rarely have a single author, you could not keep up with ranking and funding if you would do that), we did not look which journal is the best fit for our paper.

Instead, we compiled a list of journals that would potentially publish our paper given its subject matter and the importance of its contribution to the advancement of science. Then we sorted the list by impact factor and submitted the manuscript to the New Journal of Physics, with impact factor 3.264. We are obviously very objective and our manuscript will go through a second round after revision, so we picked the correct journal.

Being at the end of my tour of duty as associate editor for a journal with impact factor 0.757, I am very aware of the current bad trend of shopping one's manuscript. In this trend a manuscript is submitted to a top tier journal like Nature (28.751) or Science (26.372) and then submitted again and again to journals with lower impact factor until it is no longer rejected.

Not only does this tax the system by requiring many unnecessary reviews (handling a rejected paper costs about $500, which are not offset by page charges and download fees), but it backfires when a reviewer gets the same manuscript though a different journal. In fact, a good associate editor will seek reviewers that are most familiar with the research described in the paper, so two different associate editors from two different journals will draw on the same small set of potential reviewers. The manuscript then gets rejected as a resubmission of an old manuscript.

It is important to understand that the impact factor is a relative quantity. For example, the Cancer Journal for Clinicians has an impact factor of 69.026, but this does not mean that at 0.757 my journal sucks. On the contrary, compared to other journals in the field it performs quite well. For example its sister journal from the same publisher has an impact factor of 0.455 and that of an other society comes out at 0.220. These journals tend not to publish many survey papers, which inflate the impact factor.

Message 1: use the impact factor with a grain of salt.

As my wife says, even if you win the rat race, you are still just a rat. In reality, today you do not need an impact factor to make an impact. While we can dream of publishing in Science and winning the Nobel Prize (or the Judd Award for us color scientists), today we can shower our incremental knowledge onto humanity with publishing media like personal web sites and blogs. You are reading this post, proving it works!

Granted, it is very informal, but there are some easy steps up. The traditional better medium is the technical report, which can be freely downloaded from the institution's web site, like HPL-2008-109 in the case of my most recent collaboration with co-blogger Nathan Moroney.

For those who want to publish something more glossy, HP Labs is incubating MagCloud, which will satisfy all your vanity requirements. From this link your can order the same technical report, but this time printed on heavy glossy stock and laid out nicely in InDesign instead of LaTeX.

There is even a cloud service to publish the slides of your public presentations. For example, we wrote the above report in the Beamer class, which lays out the same content in a format suitable for overhead projectors (called beamers in German). You can then make available your slides on Slide Share, or even embed them in your blog, like so:

Finally, since in these difficult economic times everybody is jumping on LinkedIn, you can use a LinkedIn application to post your slides directly on your profile page.

Message 2: use informal media to give impact to your research.

Monday, October 20, 2008

Pushing needles

In my last post Now you see it, now you don't, I passed on the editor's summary of an experiment to research how the brain can combine different views of an object into a single object representation. The researchers inserted needles in the inferior temporal cortex of two monkeys and recorded neurons while the monkeys saccaded to an extra-foveal symbol while the symbol was changed during the saccade.

Reading it like that, pushing needles into the cortex sounds pretty scary, but in reality it is not. Thirty years ago a friend of mine was working for his postdoc with a troglodyte to find the pathway from the visual cortex to the arm's motoric control when a banana was presented. The troglodyte was not in discomfort and actually happily outlived Dr. Gisin, who died at a young age from a brain tumor.

Although Dr. Gisin's team developed a relation with their subject, the problem with such experiments is that we cannot really know what the monkey is thinking. At that time, it would not have been ethically possible to use humans for the experiments. However, in the 30 years since, the medical technology has progressed so much that needles are used clinically to excite neurons in patients with acute forms of diseases like Parkinson's and epilepsy.

Compared to exciting neurons, just recording from them is a minor invasion. Thus data from experiments piggy-backing on clinical procedures are starting to become available. The quantum leap with human subjects is that they can describe what they are thinking.

In Internally Generated Reactivation of Single Neurons in Human Hippocampus During Free Recall, published in Science 3 October 2008: Vol. 322. no. 5898, pp. 96 - 101, a team from UCLA and the Weizman Institute report on recordings from epileptic patients during clinical procedures. In short, they were able to determine that in the hippocampus the neurons are reactivated in the same pattern when a person sees a scene and when the person recalls the scene.

Thus, when we reason on the model of what is happening in the world outside our body, we are in the same state as when we actively scan and analyze the outside world. We cannot immediately discriminate between what we see and what we recall. This is the basis for such effects as memory color.

This is why in general, preferred color reproduction yields more pleasing images than colorimetric color reproduction. And this is also one of the things that makes color science such a challenging discipline.

Here is the Science editor's summary of the report:

The neural correlates of remembering can only be studied with complete confidence in humans, because the subjects can verbally report their internal experience. Brain surgery in which therapeutic electrodes are implanted in the brain of patients with intractable epilepsy provides an opportunity for doing such studies. Gelbard-Sagiv et al. report that neurons in and near the hippocampus of these patients showed specific patterns of activation for each episode of the television show The Simpsons. Later, when these same episodes were brought to mind by free recollection, the same pattern of neural activity was seen, demonstrating that, at least in the hippocampus, recall of a stimulus is accompanied by activation of the same neurons that were activated during the initial experience.

Thursday, October 16, 2008

Now you see it, now you don't

Each object can cast many different images on the eye. How can the brain combine different views of an object into a single object representation? Neurons at the inferior temporal cortex (brain area IT), the top processing level of the visual system, signal the presence of individual objects even if those objects appear in different positions.

Nuo Li and James J. DiCarlo in a report in Science Vol. 321, pp.1502-1507 recorded neuronal responses in area IT of two monkeys to different objects presented at the central position and 3 degrees above or below. By systematically swapping object identity between two objects whenever the monkey made a fast eye movement (saccade) to one particular position in the visual field, the response of the IT neuron became less selective to the objects at the swap position or even inverted its selectivity. Thus, object representations in area IT can change in a short period of time.

Wednesday, October 15, 2008

Have scanner, will travel

Recently, our reader Dan B. from Google sent a link to an article about Prof. Kent A. Kiehl he read in Science magazine Vol. 321 pp. 1284-1286 (also available here). Prof. Kiehl did his graduate work with Prof. Robert Hare at the University of British Columbia in Vancouver, whom you know from his book Snakes in Suits I reviewed over a year ago.

Actually, in that post I displayed this image, which is from Prof. Kiehl's work.

Limbic abnormalities in affective processing by criminal psychopaths as revealed by functional magnetic resonance imaging

In that post I wrote jokingly that we color scientists have an advantage over the general population because in case of a suspicious colleague we could use a modified version of the color naming experiment to unmask them, or we could walk down the hall to a colleague with an fMRI machine and collaborate on an experiment.

The Science magazine article describes how Prof. Kiehl is actually doing just that. With a custom-built mobile functional magnetic resonance imaging (fMRI) scanner — roughly $2.3 million of equipment packed into a 15-meter-long trailer — and permission from the New Mexico governor to work in all 12 state prisons, Kiehl aims to scan 1000 inmates a year.

Kiehl's research is funded by four R01 grants from the National Institutes of Health, which pay about $900,000 a year in direct costs; the Mind Research Network (MRN) at the University of New Mexico in Albuquerque paid for the scanner.

The reason for recruiting the subjects in the state prisons is that these are places where it is easy to find them. However, if Prof. Kiehl would start his research project today, he might have chosen a different venue and population. In fact, In Snakes in Suits Prof. Hare and Dr. Bobiak describe in detail the evolution to the transitioning organization and present the concept of entrepreneurial pretenders.

The book cited several instances of such entrepreneurial pretenders in the saving & loans crisis and Enron. If one follows the news, one might suspect Manhattan is a biotope where Prof. Kiehl might find an abundance of subjects to study.

Speaking about the news from Manhattan, in an earlier post I blogged about how the end of communism and the cold war made the concept of outbraining the communists obsolete and started the demise of scientific think tanks. If one believes the news, one might think we are currently witnessing the end of capitalism, with the widespread nationalization of financial institutions.

This might bring a new era where it is no longer about being number one, but just about working and making a contribution to society. And this might not require think tanks, in the view of our leaders. In fact, the latest issue of Science magazine has a news item by Laura Margottini on Italy Restricts Academic Hires, in which it is reported that academic hiring is being severely cut back and institutions should look for private sponsors. Apparently it will not be the western hemisphere who will dig world economics out of the current hole with new science and technology. We will have to look elsewhere.