We need a model for technology embedded arguments
When we describe behavior change via technology, we are describing a designer’s intention being argued via technology. The designer wants the user of a system to change his perception about the world and ultimately his behavior. The change is achieved by designing the system in a manner that would persuade the user to actually change. Currently we don’t have a model for analyzing such technology embedded arguments and furthermore it’s hard to see how we might shape such a model.
Arguments persuade - what are arguments in persuasive systems?
Most persuasive systems focus on the relationship between man and machine, where the machine is undertaking the role of persuading the user of a certain behavior. Typically systems are deliberately crafted in a manner that would lead the user to adopt this target behavior, so that they may act as the designer’s vehicle of persuasion.
This closely resembles traditional communication. The ‘designer’ of a speech deliberately crafts it to persuade the audience. When giving a speech or writing an article, we embed arguments into language itself by formulating and saying the actual arguments that the audience hears. As we look at persuasive systems, it becomes harder to determine where the arguments actually reside and even harder to analyze them.
Embedded arguments: Driving persuasion in software
Redstöm (2006) was the first to point out, that we may perceive persuasive systems as systems where a designer has deliberately embedded an argument into the system itself. As systems naturally have no intention themselves the intention is infused into the system by the designer. His intentions are built into the system itself in the form of embedded arguments. The system in turn acts as a vehicle of persuasion by driving the designer’s intentions to the end user.
An argument models for textual arguments
In writing and speech we have various models for arguments that aim at explaining what an argument really is. Such models at least offer a way to analyze existing arguments or to test them before presenting them to a target audience. One such model is Toulmin’s model (Foss, Foss & Trap, 1985) as seen below:
Figure 1: Example of Toulmin’s model used on an argument for smoking cessation
The general concept of this model is that the ‘warrant’ allows for audience to subscribe to the ‘claim’ of the speaker, by connecting the ‘claim’ and ‘grounds’. We pose a ‘claim’, and then we support it by a ‘warrant’ that would allow the listener to actually agree with us based on the ‘grounds’ we present - in essence resulting in persuasion.
‘Backing’, ‘modality’ and ‘rebuttal’ are optional elements that are often times omitted, they concern the level of certainty that we put behind our claim. Notably a warrant may be both: logos (appeal to reason), ethos (appeal to authority) or pathos (appeal to emotions). Using the model we may even find whole hierarchies of arguments within a text. The example given in figure 1 above might easily be extended. For instance the grounds “Smoking causes cancer” may in itself be seen as a claim in a different argument. This claim has its own ‘warrant’ and ‘grounds’ etc. etc.
It’s unclear how an analysis framework for embedded arguments should be formulated
Today it’s not possible to analyze arguments embedded into software in the same clear manner as it is with texts (figure 1.). While persuasive systems mature we have no models for analyzing the arguments that are embedded in them. No models allow for us to interpret the validity of such arguments. We may naturally use Toulmin’s model for analyzing the textual content that a system presents to its user in the form of: text, video or audio. Persuasive systems however employ a myriad of techniques to persuade for instance: interaction, kairos (~ timing), persistence, tailoring, personalization and reduction to mention a few.
The mix of content and interaction makes it very difficult to clearly analyze, why one embedded argument might be more persuasive than another. Why is one health application more successful in motivating users to be more active runners? What argument did the systems really pose? How could we compare the arguments posed by one application to the arguments posed by another? We simply don’t have models for analyzing those types of embedded arguments.
Conclusion
This short article has explored the notion of technology embedded arguments in persuasive software and that embedding of arguments is an elaborate part of designing persuasive systems. Whereas we do have models for analyzing classical arguments as presented in text or speech (exemplified by Toulmin’s model), there is a lack of analysis models for arguments that are embedded into persuasive systems. The complexity of persuasive systems leads us to assume that an analysis model for embedded arguments would also be more complex. Persuasive systems might contain text, video and audio, thus having the same capabilities as traditional media, but they extend far beyond the textual by employing a myriad of interaction strategies. Currently it’s not possible to analyze this added complexity of technology embedded arguments in the same manner as we could if we were for instance analyzing two texts on the same topic.
Key reading(s):
Foss, S., K., Foss, K., A., Trapp, R. (1985), “Contemporary Perspectives on Rhetoric”, Waveland Press Inc., Prospect Heights, Illinois, ISSN/ISBN: 0-88133-542-8
See also:
Toulmin, S. (1969). The Uses of Argument, Cambridge, England: Cambridge University Press.