David Souto - Cabinet of Curiosities (blog)

A remedy to the file drawer problem

Tue, 10 Apr 2012 13:07:18 -0800

Replication is a necessity in any field of science. Uncontrolled variables, sheer chance or technical mistakes can lead to false positives. The problem with replication is that they are hard to publish, especially if one reports a null effect. Related to this is the fact that we often base decisions to publish a study on statistical significance, generating a publication bias well known in meta-analysis. Studies reporting effects that do not reach the conventional level of statistical significance are likely to end in a file drawer instead of being published: the 'file drawer problem'. Not only this can inflate the importance of an effect, but it entails considerable research costs - the cost of chasing effects that others have found hard to replicate but did never have the chance to enter the scientific record. In a recent paper researchers (mostly faculty in psychology departments) reported an average of 4 fully replicated studies over 6 attempts on average.

To address this issue PsychFileDrawer.org, currently in the beta version, proposes an easy way to store information about replications. Although the website is no substitute to peer-reviewed research, it gives the opportunity to connect with researchers interested in the solidity of an effect or to evaluate it directly. Hartshorne and Schachner go even further by proposing to link information about replications to the original papers. According to them, this could shift the current (pernicious) emphasis on biometric information by allowing to evaluate the replicability of a study after publication.

The launch of the website follows a series of papers raising methodological concerns in the field of cognitive neuroscience and experimental psychology. However, nothing puts it more clearly than John Ioannidis paper entitled "Why most published research findings are false". That leaves at least a 50% margin for improvement.

Felice Varini

Sun, 10 Jul 2011 03:08:09 -0800

I stumbled recently upon the work of the Swiss artist Felice Varini, whose work is mainly scattered in France (you may find one near your area by looking at his portfolio in Google maps). His art is especially interesting among the various popular variations of the trompe l'oeil. In paintings, the artist uses principally the rules of perspective to create the illusion of a three-dimensional subject in the world (as with 'Escaping criticism' by Pere Borrell del Caso), which is necessarily optimal for a particular point of view.

'Escaping criticism' , 1874, by Pere Borrell del Caso,

Varini depicts often simple geometric shapes that come into a bi-dimensional 'gestalt' when viewed from a single viewpoint. Then, surfaces that in reality extend in three dimensions, are perceptually flattened, which is quite the opposite effect from what we expect of a trompe l'oeil. Because from this particular point of view the shape has no belonging to any object out there, they lie in a different layer than their medium, as if they originate in the eye of the beholder or as strangely palpable projections.
Some of the forms are as simple as a line that runs across the four walls of a plaza, cutting the visual space in two when viewed from its center; or a simple ellipse shown below.

'Elisse rosa piena' by Nielle Torini and Felice Varini. Lugano, Switzerland.

Some of his work takes extraordinary proportions, as in Saint-Etienne, in France, shown below. He applies the usual stencil projection technique to paint concentric circles. As a result the buildings appear to warp around the center of the composition.

Felice Varini, Saint Etienne, 2005.

Coherence of garbage

Tue, 08 Jun 2010 06:46:22 -0800

"The job of any sensory system is to create objects in the world out of the incoming proximal stimulus energy” (Stephen Handel in Perceptual Coherence, OUP).

Most of the time objects are correctly recognized without effort, however artful arrangements can reveal the fact that many solutions can arise from the same projection on the retina (the "proximal stimulus energy"). The London artists Tim Noble and Sue Webster made quite an impressive demonstration of this principle by creating a huge gap between the projection and the object that generates it (cf. image below). Even if no different from ombromanie, the art of creating images from hand shadows, it remains impressive.

When the heap of garbage is illuminated from one side we see a couple back to back, where there should only be some accidental organization.
At first I thought of it as a rather pessimistic version of Plato's allegory of the cave. The intention was clearly different though. The rubbish was produced by the artists consumption, who are thus doubly reflected by it.

Dirty White Trash by Tim Noble and Sue Webster. Exhibited in 1998.

Limits of perception

Fri, 12 Mar 2010 03:33:20 -0800

In a recent paper published in Trends in Cognitive Science, Alex Holcombe reminds us that the temporal limits of perception do not allow for instance to know whether at any moment all hooves of a galloping horse raise above the ground.

I found this hardly believable, however a painting of Gustave Courbet, exhibited at the Neue Pinakothek of Munich, indicates that this is the case. Even a careful observer - a figurative painter - can get it wrong. The painting shows a "bolting" horse with the four hooves above the ground, which apparently can never happen with the legs stretched. Courbet was a victim of the poor temporal resolution of his visual sense.

Gustave Courbet, Bolting Horse

The first analysis of a galloping horse motion is due to Eadweard Muybridge, in 1877, at the end of Courbet's life. Actually it seems that he did it in part because it was a source of debate at the time (http://en.wikipedia.org/wiki/Muybridge).

Muybridge racing horse

LondonFest

Wed, 10 Feb 2010 14:02:57 -0800

Below is a short presentation of my latest work on the "color-phi" phenomenon that took place during an informal gathering of London and Paris researchers organized by Patrick Cavanagh (http://cavlab.net/).

Watch live streaming video from londonfest at livestream.com

Close up

Tue, 27 Oct 2009 16:35:27 -0800

This website features a detail of a Chuck Close portrait. You may not have noticed at first sight that the banner is part of a face. In this kind of woodcut portraits you only see the three-dimensional shape that is represented at some distance. From too close (no play of words there) the details become inconsistent with the assumptions that the visual system makes about what should be a shadow, according to the interpretation given by Patrick Cavanagh and John M. Kennedy in Science.

Emma - Chuck Close

Chicken strengths

Wed, 22 Jul 2009 06:07:06 -0800

Chicken have a bad reputation, which I would like to restore in my first post. It is well known that they are surprisingly bad at adapting their picking when their vision is shifted by prismatic goggles (yes, chicken prismatic goggles do exist). In fact, they are so bad at it that they may starve before finding a way to correct their ballistic head movements to feed themselves.

However, there is one thing they do impressively well, namely to keep their head stable while moving their body (http://www.youtube.com/watch?v=we7OmwuOHzI). This may be useful when they walk, keeping the head steady most of the time, and then briskly aligning it with the body [1].

[1] Carpenter, RHS (1988). Movements of the Eyes. London: Pion.