« Is Ambient Technology the Opposite of Virtual Reality? | Main | Tell me if this would creep you out? »

April 16, 2006

How do people figure it out?

Mathmos Airswitch Mk1I am working on a study that asks participants to figure out how to use a device that offers no affordances or clues as to what it actually does, and looks nothing like what it actually is. The device I am using is the Mathmos AirSwitch 1. It is a lamp, shaped like an erlenmeyer flask, and is operated via an infra-red sensor that is hidden inside. I've now conducted 5 separate sessions where I had 2 participants sit together and attempt to simply figure out what it is, and how it works.

The reason for this study is that I'm trying to understand how people figure out how to interact with devices when they have no examples to compare them with. Of course this is because in ambient displays and information devices, one of the main challenges is letting people know how to interact with devices that are designed to blend into their surroundings.

After viewing the video footage collected during the experiments, I've started work with Erik on a model which may help explain how people figure out how to use devices in general. After all, in order to understand how people could figure out ambient devices (which are designed to somewhat conceal themselves), I will need to know how they figure out devices designed to explicitly reveal their functions. For example, how does someone figure out how to operate a set of computer speakers that they have never seen before, or a new mp3 player, or a novelty cigarette-lighter?

Just from everyday observations, I believe that we store abstract notions of the devices we interact with in our heads. For example, we have a notion of "computer speaker" or of "cigarette-lighter", which is formed from all the computer speakers or cigarette-lighters we've ever seen. Also, we have abstract notions of interactions that belong to each of these abstract devices. For example, cigarette-lighter may have associations with the abstract "flick" or "push". So when someone presents us with a novel cigarette-lighter, we create a mental matrix of every cigarette-lighter we've seen, along with every interaction we've observed involving cigarette-lighters. If we are lucky, the lighter presented matches one of the combinations we came up with, in which case we can predict the new lighter's operation.

I thought of this example after noticing some cheap cigarette-lighter's for sell at a gas station. They were made to look like the expensive Zippo brand lighters, but when flipped open they revealed electric coils where the wick should be. These coils heated up and turned red when the button was pressed, which is similar to how the cigarette lighters in cars function. I already knew that the device purported to be a cigarette-lighter, and I've seen both a Zippo and a car cigarette-lighters, so it was easy to deduce a hybrid of the two. One pushes the button, waits for the coils to turn red, and then pushes the end of the cigarette to the coils as you do with the car version. Other examples I've noticed involve the hybridization of a cigarette-lighter with a digital camera, and a cigarette-ligter with a pig.

If this model is accurate, then how do we figure out devices that do not have any similarity with the set of abstract devices we already understand? This leads to the research question for this study, "How do people figure out the operation of devices when the devices are completely novel?" In the Mathmos lamp study, the most interesting part of the footage may be in the short amount of time before the participants realized that it was actually a lamp. Unfortunately, that time was very short because the first thing that several of the participants did was pick up the device and look inside at the light-bulb. Once the participants had decided that the device was a lamp, they quickly started forming their map of what its possible functions could be and how to evoke them (in this case, on/off, and brightness). However, in this case none of the participants had the necessary interactions associated with their abstract notion of "lamp" in their pool of abstract interactions. They instantly started looking for a knob, switch, or button. When none of the usual interaction mechanisms for "lamp" were found, the participants began a very erratic search to fill in the gaps. The participants blew on the lamp, tapped the lamp, and came up with interesting theories, like that the rings on their fingers were causing the lamp to turn on or off.

I guess what I have to do now is design a study to test my assumption about how we rationalize the use of novel devices. I think the method would involve providing groups with two unrelated devices, and then a third that was a hybrid of the other two. If I can show that I can provide the abstract notions necessary for figuring out the third device, then I believe that will support the model.

Posted by whazlewo at April 16, 2006 7:34 PM

Comments

Hmm...you don't think we actually create these "mental matrices" on the fly, do you? I mean, it would be pretty difficult in terms of efficient computation to create an abstract model every time we saw a new example of a device...I think it's more likely that we have implicit abstract models that we reference and add to, but the examples that make them up are rarely compared to one another, especially in the form of a matrix...our brains just don't work like that.

Posted by: Josh at April 16, 2006 10:46 PM

If I understand you correctly, your theory is that when we encounter a new device, we figure out what category it's in, and we look in our catalog of experiences for interactions we've had with things in that category, and see what sticks.

Then you say you want to know how we figure out how to use things we can't categorize? How is your theory applicable in that case? If you don't know what something is, you have no "matrix" to search.

Posted by: Erik at April 16, 2006 10:56 PM

Yeah, I agree, I have no real idea how the mind/brian actually works. But I’m just trying to create a model. Perhaps we can think of things in the terms I describe anyhow. The model would say that when people see a new device they first try to figure out "what" it is, and then they try to predict how it works based on the interactions they have associated with those sorts of things. Put simply, the model would imply that we are more likely to be able to figure out devices if they are similar to other devices that we already know of. So with the example of the lighter, one opens the top and is surprised to see the heating coils when they expected to see a flame. Nonetheless, most are able to figure it out. Would they have figured it out as quickly if they did not already know how car cigarette lighters worked? Knowing about the Zippo and the Car lighter would mean that the bastard Zippo-like lighter I saw was already in the matrix, and therefore likely to be easily understood.

But you are right, I’m sure a matrix is not actually formed in the brain, but determining that matrix (and the likelihood that a person would have your device I their matrix) may help predict the difficulty involved in figuring out the device's functionality. Maybe a weighting system could be derived with some sort of distance metric. For example, I could say “I predict that Erik’s device will be 20% easier to figure out than Josh’s based on commonly known devices, but that’s a big maybe.

As for Erik, it’s not that I want to know how we use things we cannot categorize. I want to know how we add things to our catalog. I cannot remember how I figured out how to use a remote control, a desk lamp, or an iPod, but I do seem to know. Also, I'm not sure why I know how to use my new can opener, it looks nothing like my old one, and it cuts around the outside of the can vs. the top of the can, but somehow I figured it out on the first try. I know that it is a testament to the design of the opener, but development, design, user-testing, etc., only explains "how" it became a good design, not "why" it is a good design.

As a counter example, in the lamp-study, even after the participants had realized that the device was a “lamp”, it had none of the interactions they expected. The AirSwitch Lamp did not exist I their matrix of possibilities.


These ideas are fresh off the press though, I have no idea at this stage if they’ll fly ☺. Thanks for the comments.

Posted by: Richie at April 16, 2006 11:44 PM

I see some similarities between this research question and the concept of 'schema' in social psychology. Studies in this area reflect on how we carry scripts in our minds of how to behave in different social situations or environments. was thinking that you might find some inspiration here. There might be parallels between how we decide to behave in a certain environment, and how we decide to interact with a tool.

Posted by: webfoot at August 26, 2006 3:02 PM

I figured there was something to help explain what I'm talking about in either Psychology or the Cognitive Sciences. My advisor had mentioned someone in IxD or HCI who had written on schemas, but the name escapes me. Now that I think about it I hope its in my email somewhere. Thanks for the comment though, whoever you are :-).

Posted by: Richie at August 26, 2006 3:12 PM

Post a comment



Remember Me?


—>