Reading #4:  Measuring Users’ Experience of Agency in their own Actions

A study of how it feels to be a button

Introduction

This is my first blog about a paper where research was performed just for the sake of knowledge.  Every other time, there was some sort of product that they were designing and testing.  In this case, the study sought to explain how responsible humans felt for their actions in various scenarios, and then how a computer could help without removing that feeling of responsibility.  The researchers were:

David Coyle, Cambridge University, is a Marie Curie post-doctoral research fellow who's research tends to be a blend of computer and neuroscience.  http://www.neuroscience.cam.ac.uk/directory/profile.php?david.coyle

James Moore, University of London, works in the department of psychology as a lecturer.  His research involves the study of agency, and disorders that remove that feeling of agency.  http://www.gold.ac.uk/psychology/staff/academicstaff/moore/

Per Ola Kristensson, University of St Andrews, is a lecturer in HCI in the department of computer science.  His research intersects HCI and AI.  http://sachi.cs.st-andrews.ac.uk/people/faculty/per-ola-kristensson/

Paul C. Fletcher, BCNI, is a researcher who primarily deals with understanding psychosis.  I see no indication that he would have even been interesting in this study...http://www.neuroscience.cam.ac.uk/directory/profile.php?pcf22

Alan F. Blackwell, University of Cambridge, is a researcher who works in the Computer Laboratory at Cambridge, and whose website is fairly difficult to navigate.  http://www.cl.cam.ac.uk/~afb21/

Summary

Agency is the feeling of responsibility a person has for their actions.  Agency is important because it controls how a person feels about the existence of their own free will.  The study sought to test how that sense of agency is affected in two cases:
1.  When the user treats their skin as the user interface, instead of a touch screen.
2.  When a computer program helps the user to interact with objects with a mouse on a normal computer screen

Related Work

Paper about how changes in the workstation a person is used to hampers their sense of agency in their work:  http://www.jstor.org/stable/10.2307/248811
A study on how humans feel about and interact with machines:  http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.109.8074&rep=rep1&type=pdf

Kind of the basis of this research; can a program be made to effectively reduce the workload for a user:  http://portal.surrey.ac.uk/pls/portal/docs/PAGE/COMPUTING/PEOPLE/N.ANTONOPOULOS/TEACHING/CSM13%20SOFTWARE%20AGENTS/CSM13%20COURSEWORK%202006/PAPER2.PDF
The next step after this paper; how can we effectively utilize this knowledge to make an interface agent work well (book):  http://books.google.com/books?hl=en&lr=&id=TNl1Eb4LgmsC&oi=fnd&pg=PA111&dq
=Autonomous+Interface+Agents&ots=YtfyGPGa4s&sig=MrHIuPbVxYIk-gh76_b_V_mIB3o#v=onepage&q=Autonomous%20Interface%20Agents&f=false
Discusses how a sense of "presence" affects immersion and agency within a program: http://www.mitpressjournals.org/doi/abs/10.1162/105474601300343603
Explains the concept of agency:  http://www.sciencedirect.com/science/article/pii/S1053810003000527
More on agency and its importance:  http://www.sciencedirect.com/science/article/pii/S0010027708002382
How agency is influenced by external sources:  http://www.sciencedirect.com/science/article/pii/S1053810009001020
About the concept of will:  http://www.sciencedirect.com/science/article/pii/S1364661303000020
How computers have affected the agency of humans:  http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.201.5998&rep=rep1&type=pdf

Evaluation

The first test in this paper had half the users tap on their skin, and then estimate how long it took for a light to turn on after tapping the "skin button" on their wrist.  This is called "binding time".  The other half pressed a physical button and made a similar estimate.  The actual time it took was measured electronically, so that they would have an objective time.  The purpose of this subjective quantitative data is to determine a user's agency.  If binding time is greater, then they experienced greater agency than if the binding time was less.  Generally, users felt greater agency when tapping on their wrists.

The second test had users click on a button on a computer screen and again estimate the binding time.  The four groups had varying amounts of computer help (by giving the button "gravity"), from none to ridiculous amounts of help.  The results showed that users felt greater agency with a small amount of computer assistance, which sharply decreased as computer assistance increased.

Discussion

The stated effects of this study are to provide designers with new ways of refining interaction techniques and interfaces so that users experience an increased sense of control over their actions.
In short, the study sought to have user interfaces help users interact in such a way that they do not realize they are being helped, so that they still feel completely in control.  This seems familiar...

Oh, right