Dr. Marco Kalz » Read

Feedback loops and the new behaviourism

Marco Kalz — Wed, 20 Jul 2011 18:00:18 +0000

The recent Wired article about feedback loops by Thomas Goetze has helped me to understand the phenomenon of sensors, feedback, learning and behaviour change better. Or better to say, after the article I was aware how pervasive these loops are already part of our lifes.

New sensors and feedback via personal devices have inspired many developers and companies to use live data measures for intelligent information distribution in many parts of our lives (energy consumption, sports activities, health awareness etc.). What surprised me about the obvious success of these feedback loops is their simplicity. In its core they often realize a behavioristic stimulus-response model and to some extent also conditioning effects. Dan Lockton (via Stephen Downes) has written an interesting article on his blog where he discusses the implications of behaviourism for designers. He discusses the different conditioning types and he presents an example of a simple feedback loop:

"Pryor (2002) gives the example of a car seatbelt warning buzzer as negative reinforcement—a device designed to be irritating or unpleasant enough to cause the user to take action to avoid it. We might consider that a recorded voice saying “Thank you” after the seatbelt is fastened could be a positive reinforcement alternative. Positive and negative punishment are essentially the inverse of each of these—a fine for not wearing a seatbelt while driving is a form of positive punishment, and taking away someone’s driving licence would be a form of negative punishment" (Lockton, 2011).

On the other hand, many feedback loops (and this is what Goetze has neglected to some extent) also seem to work because of social control or social competition effects. Let's take the famous example of running trackers and sensors. People use individual sensor to control their speed and perseverance and after their running session they share their efforts with a specific audience (the family, facebook friends or the world). This again seems to have positive effects on motivation and in the end healthiness. But there can also be a negative effect of the social embeddedness which Lockton calls social fences or social traps. A social trap is a situation where reward and negative consequences are given but they overlap each other. Lockton points to a social trap taxonomy by Cross and Guyer (1980) which discusses 5 classes of traps:

time-delay traps (delay between behaviour and reinforcement is too long to be effective)
ignorance traps (knowledge about negative effects is not used because the rewards establishes a habit)
externality traps (negative effects are not on the individual but on other people)
collective traps (similar to the tragedy of commons)
hybrid traps (a mixture of some of these traps)

For the design of ambient displays and feedback loops it would be very interesting if we could break these traps and connect behaviour and consequences directly, especially with a focus on the grand challenges of our time (e.g. sustainability, security etc.).

Contextualized Learning and the fireplace

Marco Kalz — Wed, 28 Apr 2010 14:16:35 +0000

photo credit: The Lightwor

David Wiley once coined the sentence that content is only the fireplace that we use to gather around. One could continue that learning does not happen via the burning wood alone but learning is always happening in a given (social) context that influences the learning process. For mobile learning the topic of “contextualized learning” is a very important one since the context of learners has to be taken into account for the design of innovative mobile learning scenarios. But how can we approach the complex topic of the context of a learner?

Figueiredo & Alfonso (2006) develop a philosophical framework to describe the relation between context and learning. In this framework positivist and constructivist perspectives on context are compared and discussed. According to the authors context is viewed from a positivist perspective as being part of a realist world and therefore it is an entity external to the learner. It is often defined as the environment around a learning event. This leads to the assumption that context is a predictable variable that can be controlled in advance. From a constructivist perspective “context cannot be located and delimited…[and] context is only perceived through its interaction with the learner”. Most of the papers about mobile learning follow a positivist view of context. In this view context variables can be changed, controlled and compared in (quasi-)experimental setups and the influence of contextual variables can be compared. Due to the high usage rates of mobile devices and the increase of mobile internet flatrates the context of learners is more and more an authentic one that cannot be easily controlled through lab experiments. Therefore the constructivist view of context will get more attention. At the same time intelligent research designs are needed to conduct research about the influence of context to learning and the development of context aware learning support services.

[Disclaimer: This posting is part of a series of the current hot topic "mobile learning" of the Learning & Technology Community]

Foundations of a new science of learning

Marco Kalz — Tue, 27 Apr 2010 20:08:43 +0000

During the ECTEL-Conference last year Mike Sharples had pointed to an interesting paper which has been published in the highly recognized journal Science. Due to the fact that this journal does not publish often articles related to Technology Enhanced Learning I was curious what the paper “Foundations for a new Science of Learning” is about.

photo credit: epSos.de

The paper proposes a new science of learning based on 4 contributing disciplines: Psychology, Education, Machine Learning and Neuroscience. The authors stress the aspect of social learning of children on the one hand, on the other hand they discuss the similarity between machine learning research and learning processes of children (e.g. pattern recognition, covariation or frequency distributions).

While psychology, neuroscience and machine learning contribute the hard facts to the new science of learning education is mostly seen interesting because of three activities: Early intervention programs for their so called “cascading effects”, non-formal learning approaches and classical 1-to-1 face to face teaching. For the last aspect the use of intelligent tutoring systems is mentioned as a promising research direction. The article ends with the following sentence:

A convergence of discoveries in psychology, neuroscience, and machine learning has resulted in principles of human learning that are leading to changes in educational theory and the design of learning environments.

The article collects a wide range of interesting findings from the disciplines mentioned above. On the other hand it left me with unease since the authors must be aware that there must be more then just a ‘convergence of discoveries’ to build a new science of learning. Also their account that “new learning technologies are being developed that embody key elements of individual human tutoring while avoiding its extraordinary financial costs” was surprising since I did not yet realize that ITS have become a cheap activity. Anyway, an interesting paper with many current and important findings and references.