Models have many functions in science, but two of the most important ones are that they synthesize what is known, and that they allow researchers to derive new predictions. Both functions have been central themes in my research to date. My work is built around computational models of brain and behaviour. In these models, I integrate anatomical and physiological knowledge with detailed behavioural analyses. I apply the models to get at the central question of how the brain produces behaviour, and also to get insights into abnormal states of the brain: the damaged brain, the brain in psychiatric disorders, and under psychoactive medication.
Computational modelling is not an independent branch of science – its goal is not solely to build beautiful theories but to derive new hypotheses for empirical scientists to test. In my own work, this means that I strive to create models that produce precise predictions at the behavioural, pharmacological, neuropsychological, and neurophysiological level. This makes my models a platform of multidisciplinary integration. To test model predictions, I have set up my own lines of empirical research, and have developed collaborations centred on behavioural research, patient research, neurophysiology, neuroimaging, and pharmacological research in humans and animals.
One line in my research involves reading. I have developed a model with Jonathan Grainger (OB1) that integrates the literature on reading single words with the literature on reading texts (a plausible reason for why these two weren't integrated to begin with can be found here). This model is currently being extended in two directions: reading aloud so that learning to read can be modeled, and semantics.
Since my PhD, for many years I focused on learning and memory. One line of work looks at how memories are bound to context. With Lucia Talamini (UvA), I investigated the idea that abnormal processing of contextual information plays an important role in schizophrenia . Our idea is that these deficits are primarily the result of abnormalities in the parahippocampal gyrus (the area next to the hippocampus). This brain area, next to the hippocampus, integrates input from all areas of the brain and sends it on to the hippocampus. In patients with schizophrenia, this integration does not work as well as in other people. In particular, contextual information is not integrated very well with information about the items in the focus of attention. This, we suggested, may underlie many symptoms of schizophrenia. In another collaboration, with Vanessa van Ast (UvA), I look at the role of context in the generation of pathological fear. The idea is that traumatic experiences become particularly disrupting when they were not linked well to their original context.
A second research line, funded by a Vidi grant related to how our brain responds to new stimuli. This project, is described here.
Much of what we learn is not easily verbalized - it is what's called nondeclarative. I have looked at nondeclarative learning in several projects, often with an eye on testing whether such learning is really that different from consciously accessible, verbal learning. With Mark Gluck (Rutgers Univ., Newark), I worked on a model in which the rodent classical conditioning literature is reinterpreted in the light of our knowledge on primate and human episodic memory. In particular, we argued that the same kind of storage is at the basis of both conditioning and episodic memory. An extension of that model to orienting towards novel stimuli inspired my project on novelty.
I have also done behavioural and electrophysiological studies on
priming in visual attention. This theme is gaining prominence
because it suggests that cognitive control not a wholly free agent
(i.e., a homunculus who decides what you do), but is partly the
slave of recent experience. The project of my PhD student Wouter
Kruijne was partly geared towards understanding priming in detail.
We have found that it dissociates into two forms, one
a short-term process that seems to be an automatic consequence of
looking at stimuli, and
the other a process akin to remembering previous trials.
Interestingly, context seems to play a similar role in the latter process as in episodic memory - again linking nondeclarative learning and episodic memory.
Olivers (VU) I created the model Boost and Bounce model of
access to working memory. This model undercuts the received view
of the attentional blink, a much-investigated
effect in which consciousness seems to ‘blink’ for half a second
after detection of a target. Our PhD student Eren Gunseli worked on how people learn to search for a target, and the role of working
memory in this process.
In my PhD research with Jaap Murre
(UvA), I focused on the loss of remote memories after brain
damage. Theoretically, such loss is interesting because it
suggests where remote memories reside in the brain. Moreover, the
work led me to investigate wit Walter
van den Broek (Erasmus University Rotterdam) memory loss after electroconvulsive therapy (ECT). ECT is often successful in combating severe depressions, but is highly controversial because
it is thought to erase parts of a patient’s remote memory. With
mathematical remote memory tests developed with Jaap Murre, we
showed that most of it is temporary. I still maintain a test for
retrograde amnesia, the AMV.
With two previous PhD students, Wouter Kruijne and Stefan
vd Stigchel I built a low-level
model of the eye movement system in the primate brain, and
systematically explored saccade trajectories as a way to uncover the processing that occurs during a saccade. My PhD student Nicci Anderson works on uncovering the role of salience in the programming of eye movements, in particular in everyday scenes.
I worked on some applied projects, most notably on noise annoyance with the NLR. This is the project of my PhD student Kim White. I have also looked at the use of cash for the Dutch National Bank.