Wetenschap - 28 augustus 2014

Measuring behaviour

Rob Ramaker

Cameras that record emotions, a heart-rate monitor that detects your preferences and apps that control crowds. Technology for measuring and influencing our behaviour is booming in the social sciences.

Twenty minutes was how long it took the fictional technology company The Circle in the novel of the same name to find any given person on Earth using public cameras. Reality has not yet caught up with science fiction but the idea no longer seems wacky. Take the brand-new PeopleTracker system, which uses a network of interconnected smart cameras to let consumer scientists record every little detail of consumers’ movements in a supermarket. 

PeopleTracker is one of the new techniques being presented at the  Measuring Behavior 2014 conference at the end of August (see below). At this conference, researchers from more than 30 countries will be presenting the latest methods for monitoring human and animal behaviour.

Innovative techniques for measuring behaviour are making headway in the social sciences in particular. For example, Erica van Herpen, assistant professor in the Marketing and Consumer Behaviour group, is helping with the further development of a virtual supermarket that the Wageningen Social Science Group has used for some years. Experimental subjects ‘walk’ through a specially designed environment that they see on three plasma screens in a semicircle. Researchers hope this will help them discover how to persuade consumers to buy meat that is more animal-friendly (and so somewhat more expensive). Van Herpen: ‘Experiments with virtual supermarkets have been shown to give better predictions of what happens in practice in various respects than old-fashioned experiments with photos of shelves.’


Social scientists are interested in direct measurements of our behaviour because what we say does not always match what we do. The answers we give in questionnaires are not a perfect prediction of what we choose in practice. For instance, sometimes we don’t have a good view of how much we eat. Or we say that we would never dream of buying factory-farmed meat because that is the socially acceptable answer. ‘When you are asked personal questions, you often give answers based on how you want to be rather than how you are,’ says Stefanie Kremer, DLO researcher at Food & Biobased Research. 

Scientists also benefit from new technology because it takes over the boring or time-consuming jobs. There is a trend in animal research, for example, towards automating observations. That has several advantages, says Andrew Spink, senior consultant at the technology company Noldus and a member of the Measuring behavior programme committee. Researchers no longer have to keep a tally by hand of how their mice behave, for instance; instead, they can have cameras record where a mouse runs, what it eats and whether it exhibits anxious behaviour. This means a researcher can collect more data. And the data is often of better quality as cameras and computers don’t get tired and are not unconsciously influenced by the desire to get a certain result. 

Spink notes that the technology for recording behaviour has become easier to use in recent years. That is one reason why the techniques are now being used in all kinds of disciplines. The rise of the smartphone has also contributed indirectly. These devices are packed with sensors that have rapidly become smaller, more energy efficient and cheaper in recent years. So researchers can spend the same as before yet perform more measurements for longer. 

One place where scientists have continuously been experimenting with new measurement methods over the past few years is the Restaurant of the Future on the Wageningen campus. It looks like a perfectly ordinary canteen but it has cameras that monitor visitors, automatically record their behaviour and combine this with cash-register data. Even more research is going on behind the scenes, says Stefanie Kremer. For example, they are investigating whether people think low-salt products taste worse than the regular versions. 

Informed consent

With all these new options, researchers still need to consider whether a measurement does actually answer a question. ‘We’re perfectly able to count how many sips and bites people take, for example,’ says Kremer. ‘But at present this is not enough for us to really do anything with the data.’ You will only be able to use it to measure food intake once you know how much people eat per bite – a fairly constant quantity. 

The designers and users of these techniques also encounter ethical issues. The Circle, the fictional company that tracked people down using camera images without their permission, may not have cared two hoots about privacy, but this is taken very seriously in the PeopleTracker, explains Spink. The system uses anonymized data and does not store any video material, which means results cannot be traced back to specific individuals. Even so, you still need ‘informed consent’, in which the experimental subjects consent to taking part after being given sufficient information. In a supermarket, you could do that using large information signs, for example. ‘There’s always a trade-off,’ says Spink, ‘between how much you tell people and how much you influence their behaviour.’ After all, people’s behaviour is affected by the knowledge that they are in an experiment. 


Measuring behavior 2014

Measuring Behavior is the title of an annual conference focusing on technology for behavioural research. The ninth edition is taking place in Wagenin-gen from 27 to 29 August. It is being organized by the technology company Noldus, which was founded in 1989 by Lucas Noldus after he gained his doctorate at Wageningen University. Although Noldus is obviously a for-profit organization, the company says the conference is all about science. So competitors are welcome and the agenda is determined by a panel of independent scientists.  

Crowd control

A prototype of an app that can gauge the mood of a crowd will be presented and tested during the Measuring Behavior conference. Participants who download the app will be able to see exactly where the event’s hotspots are, with the best debates, for instance, or the most flavoursome sandwiches. Or the best beats if you were using the app at a dance festival. Visitors can also use it to avoid trouble spots with a lot of stress or even aggression. To determine this, the app measures background noise and crowd density. SWEET, as the app is called, also uses facial recognition to assess bystanders’ emotions. Meanwhile, people give their subjective opinion by filling in questionnaires. So there are lots of different inputs, but the questionnaires are the weakest link. ‘Obviously you won’t be spending time on the app if things get really interesting,’ says Andrew Spink, senior consultant at the development company Noldus.