Inhaltspezifische Aktionen

Dan Kersten

Dear Dan, Dear students, colleagues, ladies and gentlemen,

This is now the 13th time that the Kurt-Koffka medal is awarded. Kurt Koffka was a pioneer of psychology and Gestalt theory.

As most of you may know, Koffka is particularly famous for his work on perception and child development – work that partly originates from the time when Koffka was lecturing in Giessen for 16 years from 1911 to 1927.

The Kurt-Koffka medal is meant to honour scientists who shaped the fields of perception or developmental psychology to an extraordinary extent. Our winner this year, Prof. Dan Kersten from the Department of Psychology at the University of Minnesota, is without doubt one of the most influential figures in perception research.  It is a great and special honor for all of us that you, Dan, are prepared to accept the Kurt-Koffka award from our faculty. And, it is an enormous pleasure for me to highlight a few of Dan’s many great achievements today.

But let’s first take a moment to look at how Prof. Kersten’s career in science started.

Young Dan’s passion for understanding physical and biological patterns was awakened and fostered by countless Saturday afternoons full of experimentation with dangerous chemicals, - and expeditions with his father into the woods of South America. From the woods - he went pretty much straight into MIT, Boston and graduated, in 1976, from one of the world’s best engineering programs, with a Bachelor of Science in Mathematics.

There, at MIT, listening to lectures by the likes of Whitman Richards, Hans-Lukas Teuber and Richard Held, he decided that a future in brain science would be right for someone as mathematically-minded as him. – What a good decision that was! – From MIT he went to the University of Minnesota and right into the next world-class math program - although it was not just the science that lured him there. I suppose it was no coincidence that his lovely wife Joanne, who is also here with us today, had taken up a position as a teacher in the Twin cities.

After a Master’s degree in mathematics in 1978 Dan finally felt ready to fully take on the mysteries of the visual system and started to work with Gordon Legge towards a PhD in experimental psychology. You can tell from his first papers and the title of his doctoral dissertation - “A comparison of human and ideal performance for the detection of visual patterns“ - that Dan had come full circle connecting his love of math and natural patterns. Towards the end of his PhD - but also in the following few postdoctoral years at Cambridge, England and as an Assistant Professor at Brown University - he had encountered of what would be important and long-term companions in his work: Thomas Bayes and the ideal observer.

The ideal observer is - of course not a person (though I wish it were and I could run her in my experiments) - but a theoretical perceptual system that performs a specific perceptual task in an optimal way. It is a neat tool to develop models about perceptual processes, which can then be compared against human performance.  Some of Dan’s most influential works use such theoretical observers.

Which brings me to the contribution that Dan is probably most famous for: Like no other he saw the potential of how Bayesian inferential statistics can provide a unified framework to study perception. In his more than 1000 times cited article ”Object perception as a Bayesian inference” – together with Pascal Mamassian and Alan Yuille - he elegantly laid out how Bayesian theories can provide an account of how the visual system processes and combines large amounts of ambiguous data with prior information to yield unambiguous percepts.

Of course, he did not just theorize about it, he also put this idea - and the ideal observer - to work and showed in many seminal studies and absolutely fantastic demonstrations how visual information is integrated and how prior knowledge can affect our perception of object properties. Quite a few of these highly influential papers were published in Nature, which is probably THE most prestigious scientific journal in the world.

For example, Dan and his student David Knill, showed that the perceived shape of an object affects the visual system’s interpretation of luminance patterns in the image, specifically the perceived lightness of a surface. This was quite outrageous at the time, because up until then it was thought that lightness invokes only early sensory mechanisms that act independently of, or before, the estimation of other scene attributes.

In another compelling demonstration – that also was published in Nature - Dan, together with Marina Bloj and Anya Hurlbert, showed that the human visual system incorporates prior knowledge of the physics of light reflection between surfaces at an early stage in color perception. -- What happens in this demo is that the perceived color of a white paper - adjacent to a bright red one - changes from pale pink to deep magenta when the perceived shape of the card flips from concave to convex. Dan used an ideal observer analysis and Bayesian inference to demonstrate that this shape–color contingency arises because the human visual system must intrinsically understand the effects of mutual illumination.

Many of Dan’s publications and wonderful demonstrations are known by almost any student in psychology and are found in every textbook on perception. It is therefore not surprising that there may be no other scientist that is mentioned more frequently by name during a course in perception.

Most of you know, for example, the illusory motion of a square in depth induced by the motion ‘its’ shadow. This not only made it also into Nature because it is such a cool effect, but because it tells us something important about the visual system, namely that your brain is more likely to interpret changes in the image as being due to a movement of the object, rather than a movement of the light source.

Taken together, this line of research has been truly leading to a step change in our conceptualization and understanding of visual perception. But, Dan wouldn’t be Dan if he had restricted himself to computational modelling and psychophysical methods.

The Center for Magnetic Resonance Research at UMN is probably one of the world’s best neighbourhoods for doing neuroimaging research. So, it is not surprising that Dan couldn’t help but seeing the potential of this technique to tackle one of his favourite questions about the brain, i.e. what is the computational function of the feedforward, feedback and lateral connections that exist in the brain.

Many of the papers from this line of work that investigate cortical mechanisms in perceptual grouping, lightness or size perception, already belong to the ‘classics’ and were published in highly influential journals such a PNAS, Nature Neuroscience, or Current Biology. In these beautiful studies Dan and his colleagues - like Scott Murray, Huseyin Boyaci or Fang Fang - showed that a change in interpretation of an image also leads to changes in the neural processing of the stimulus, and provide thus -in a way - a direct glimpse at how the human brain may incorporate feedback information and prior knowledge.

The list of Dan’s outstanding contributions is long and I can’t possibly do it full justice in this short amount of time, but I think it is fair to say that Dan’s work revolutionized how research in perception has been done during the past decades.

I was lucky enough to have worked with him at some point during my career, and to have benefited from his creative way of thinking, meticulous work style, vast scientific experience, generosity, and kindness.

The Kurt-Koffka award is meant to honor scientists who advanced the fields of perceptual or developmental psychology to an extraordinary extent. My colleagues and I are deeply convinced, that you, Dan, have advanced and are still advancing the field perception to an extraordinary extent.