Just been reading an article on Why touchscreens in cars don’t work (good article, give it a read) when I came upon a theory regarding the amount of cognitive load required to utilise Voice as an interaction method and why it’s not necessarily the preferred method for interaction for a user. Well, I say a user, I’m pointing fingers at those of us closer on to dead on the alive/dead scale than those fresher from the womb. This all started with this quote from the article:
“It turns out that, as with touchscreens, using voice puts a lot of cognitive stress on the user. From thinking of what to ask for, to how to word it, to actually asking for it, the participant finds it much quicker and easier to just reach over and press a couple of buttons.”
I think this is a truism that applies to voice generally. The amount of times I verbally trip over myself starting a walk work out or even turning on my conservatory lights (mainly these lights for some reason, maybe because they are new which is interesting, is there something like vocal muscle memory?). Just because of the load of having to work out the command, then vocalising it, creates extra steps that the brain has to consider.
So I posted these thoughts off to another Neill at work, he responded thusly:
Doesn’t surprise me, given that it’s so much easier to unconsciously move my leg or switch a button than it is to think that I want to do it, work out how to vocalise it and then actually do it!
Interestingly, this also relates back to something I’ve been interested in regarding how I personally use my voice devices.
It’s typically for things which require calculation or text anyway – e.g. I set almost all of my timers, alarms and definitely all of my reminders via voice. These already require me to think in terms of words to do properly – especially the reminders, which require me to work out a full string of text which will be related back to me at a later date, so has to be intelligible – so it’s almost no extra effort to vocalise it.
I think I’m with him there. I rarely balls up a command to set an alarm or timer. By the time I’m going to set a timer, my brain has already thought “Oh, I need a 5 minute timer” (for example) at which point I don’t need to think about what the task is, my brain just needs to complete the physical steps to action it. My other examples were time sensitive to a degree (I’m walking into the conservatory and notice the lights are off or I realise I’m leaving for a dog walk without a workout going) or maybe require a fresh set of thoughts to formulate the command. Alternately the complexity of the command only has one variable (the time). My other examples require two variables (room/lights, & workout status/workout type). Maybe all 3 factors play in and that’s the sum total of the load of that task? God, I’d like to test all of this.
So the cognitive load at a high level to complete a voice task could be described as:
level of pre-emptive readiness + time sensitivity + complexity of command = cognitive load
After that has loaded into your brain, the command gets passed to the lower level part of the brain responsible for those unconscious everyday actions for implementing. Trying to access the low level implementation before the command has crystallised increases the likelihood of error on the users side.
There, that’s some sciences for you.
(FYI, Reminders are a different beast for me as ~5 years in I still can’t work out the correct syntax to get Siri to set the reminder text as I’d like it.)
But accepting this, I wonder if this could be an age thing you know? I wonder if kids today have the same issue or whether their brain will just adapt easier as they will be introduced to that way of thinking, conceptualising commands on the fly, at an earlier stage in their development. Like how learning multiple languages is easier to do when you’re a child and almost impossible for an adult, you actually have to break years of training that your brain has developed for consciously completing a task.
I put this to Neill again and he responded:
I think the adaptation is spot on, as it’s basically just learning a syntax. Similar things happen with memory – master chess players can remember incredibly detailed board setups and lists of chess plays (e.g. the ones which go like 1. e4 e5 2. Nf3 Nc6 3. Bb5) easily, but are no better or worse than other memory tasks. They can be trained to remember other things (e.g. shopping lists) using chess plays or references as a guide and suddenly their memory for those things shoots up in quality, too.
So there we go, absolutely no evidence to prove this, just a largely untestable hunch. However, a hunch backed up by observations of similar brain training having an effect. So if there wasn’t enough to fear from the generation coming through, these kids will also potentially have brains hardwired to better utilise the interaction models of the future than you do. Might as well size up that coffin now.