Multitasking – its effects are worse than you think

I’ve recently come across some surprising results out of a 2012 study on multitasking. Not only does forced multitasking badly affect performance (as you would expect) – but autonomous multitasking – choosing how you get your work done – is almost as bad.

This study has implications for how we design our own work. But it also raises questions over the conventional wisdom that it’s best to devolve as much choice as possible to the “front line”.
Let me explain.

The study

The study is described in a paper in the Experimental Economics journal (Thomas Buser and Noemi Peter, “Multitasking”, Experimental Economics, Dec 2012, Vol 15, Issue 4).
Buser and Peter set up the following experiment:

• They gave participants 12 minutes to complete a Sudoku puzzle and 12 minutes to complete a word search puzzle.
• They divided the participants into three groups.
• In Round 1 all three groups had to work sequentially. That is, they spent a full 12 minutes on the first task (Sudoku or word search – it was randomly determined which one to start on for each participant) and then a full 12 minutes on the second task (word search or Sudoku, whichever wasn’t done first).

So, so far, so good. All participants – in all three groups – have had the same work pattern on the two tasks of Sudoku and word search. Work for 12 minutes on the first task, uninterrupted. And then turn your attention for another 12 minutes to the second task, uninterrupted.

And all three groups performed basically the same (summing the scores for the two tasks together):

• Participants in Group 1 scored an average of 185 points.
• Group 2 scored an average of 191 points.
• Group 3 scored an average of 200 points.

But it’s in Round 2 that things start to get interesting.

In Round 2 the following work styles were allocated:

• Group 1 repeated the work style of Round 1. That is, they performed the first task (Sudoku, or word search) for a full uninterrupted 12 minutes and then the second task for another full uninterrupted 12 minutes. (Obviously, the puzzles were new!). These guys were called the “Single Group” – because they did a single task at a time.
• Group 2 was called the “Multi Group”. In this group the participants were forced to switch between the two tasks about every four minutes (although they didn’t quite know when they were going to be “interrupted”). The total time spent on each task was still 12 minutes – it was just split up.
• Group 3 was called the “Choice Group”. They were allowed to switch between the two tasks as often as they wanted (or not), whenever they wanted, provided the total time spent on each task came to 12 minutes at the end of the Round (they had a clever timer system going).

The surprising results

The results were surprising. Well, at least for Group 3 – but let me get to that in a second.

Firstly, as you would expect, the performance of the Multi Group (Group 2) – the group that were forced to switch between tasks – the group that kept getting “interrupted” – was badly affected:

• The Single Group’s (Group 1) results improved from Round 1 to Round 2 by 6 points (185 to 190, with rounding).
• But the Multi Group’s (Group 2) results declined by 17 points (191 to 174).
• In other words the relative decline in performance of the Multi Group was over 10% (a total of 23 points).

That’s what we all expect these days. The forced multitasking group did worse than the group that was able to focus uninterrupted on each task for the full twelve minutes.

But it’s the Choice Group (Group 3) that’s more interesting. On the face of it, you would expect that if you give someone the power to decide on how they will get their work done, their performance would improve. Remember, each participant in this group was allowed to choose how long to work on the Sudoku before going to the word search. And then they were able to choose to come back to the Sudoku with “fresh eyes” (which a lot of people reported as the reason for switching) whenever they wanted. The only restriction on participants in this group was that the total time spent on each task had to sum to twelve minutes.

Contrary to expectations, this Choice Group – the group with the most autonomy – did basically as badly as the Multi Group – the group that kept being switched around:

• Remember the Single Group’s average results had gone up by 6 points from Round 1 to Round 2.
• And the Multi Group’s results had declined by 17 points from Round 1 to Round 2.
• The Choice Group’s results also declined by 15 points from Round 1 to Round 2 (200 down to 185) – basically as bad.

So what does this all mean? Before I get to that, let’s just check that we all agree on what “multitasking” is.

What is “multitasking” anyway?

When we think about multitasking we may think that it is doing two cognitive tasks at once (I’m not talking here about things like doing the dishes while talking on the phone). For example, doing a Sudoku while at the same time thinking about what we will say to our boss tomorrow about why we haven’t finished that report.

But actually, scientists have known for a long time that we don’t actually do cognitive tasks simultaneously in our heads. We don’t both read an email and listen to the teleconference. Rather, we switch between each task – so that we are doing one task then the other and then switch back.

In other words multitasking for cognitive tasks is actually sequential work, not simultaneous work. The switch between each task may be so rapid that it feels simultaneous, but brain scans show this is not the case. (There are a number of useful references in this paper I was reading if you want to follow up – particularly footnote 6).

So what does this all mean?

So what do the results from the above experiment mean for us as we navigate modern work life?

1. Try to avoid interruptions

Clearly the first thing to say is try to structure your work so that you’re not interrupted while you are working on a demanding cognitive task. If you have to write that report for your boss make sure you switch off your phone and email and go somewhere your co-workers can’t find you.

What the results show is that multitasking carries significant “switching costs”. When you switch from one task to another there are a number of cognitive penalties you pay. In the experiment participants had to remind themselves where they had got up to last time. They had to remind themselves of the rules and the task itself. All of these carry cost. And it’s these “switching costs” that would seem to penalise performance.

So, every time you stop writing the report to take that phone call, when you come back to writing you will have to pay the “switching cost”. Better to focus for a sustained period of time.

2. Don’t assume you should switch to a new task when you feel like it

This is the biggest “takeout” from the research. Remember the performance of the Choice Group (the autonomous group) was basically as bad as the performance of the group that kept being forced to switch. So, don’t trust yourself that you know when to switch tasks. For example, if you’re writing a report and you get stuck after 15 minutes, don’t think that looking at your emails for five minutes is harmless. It will entail switching costs both as you switch to emails and as you switch back. This research would say it’s better to stay stuck for a few minutes on the report and try to regain momentum.

3. Clearly, time boxing and “sprints” still work

You’ll note that this experiment dealt with doing two tasks that totaled only 24 minutes. It has no data to shed light on sustained focus for longer than this. So common sense should prevail.

Clearly, what these results don’t say is that you should sit at your desk for hours at a time until the report is written. Rather, all the usual advice about working for 45 minutes and then taking a quick break and moving around still hold. Personally, I find the “pomodoro technique” to be fantastic if I have to get a report written for a client.

4. Don’t assume that autonomy is always the best policy

The surprising result of this research is that autonomy may not always be the best policy for workers. Maybe it’s better to be directive on some tasks than letting workers decide how they will undertake them.

Clearly more research is needed. But these results imply a few things:

• Workers left to their own devices may well switch between two good tasks (eg writing the report and responding to email) in a way that is detrimental to both tasks.
• Planning how you will undertake a series of tasks as a worker may well impose a cognitive load that lessens overall performance. Better to be told do this until it’s finished and then do the next thing.

5. Ban phones and tablets in meetings and workshops

This research just drives home to me the impact of “switching costs” on participants in workshops. Over the last few years I have taken an indulgent attitude as a facilitator to (the few) people in workshops who have felt the need to check their emails from time to time. I have assumed that we are all grown-ups and that people have the right to do this if they think it’s necessary.

After reading this research I am now less sanguine about this. Every time a participant in a workshop “multitasks” and looks at their email they have actually switched their attention. They have reduced their cognitive energy for the workshop by incurring switching costs both in the switch to their phone and then in the switch back to attending to the discussion around them. Do that a few times in a day and the costs add up – lessening the overall contribution they can make.

So, if you see a metal detector outside the door of the next workshop you attend, you’ll know what’s going on.

Multitasking is a “given” of the modern workplace

Having said all this, multitasking can’t be avoided in the modern workplace. In a collaborative environment, people will need to be able to get in touch with you. The pace of work requires it.
However, it seems to me that a good appreciation of the impact of “switching costs” is useful as we approach our days. If we can find space for some “clear air” our performance is likely to improve dramatically.