Cognitive Psychology · Simplified
Dual-Task Capacity Laboratory
Two tasks you can do without thinking. Then both at once, which is a different matter entirely.
Block 1 of 3
The letter task, on its own
A letter appears in the middle. Say whether it is a vowel or a consonant, using the two buttons or the up and down arrow keys. Nothing else is happening in this block.
What this needs. Four keys or four buttons, and about four minutes. The three blocks always run in this order, because the cost of doing both can only be measured against how fast you were doing each one alone. No trial is ever timed out.
What it cost
Two costs, and they are not the same size
The pale bar is how fast you were doing that task on its own. The dark bar is the same task while the other one was also going on. The gap between them is what the second task cost you.
All the values behind this chart
| Task | Alone | Together | Cost | Accuracy together |
|---|
Key idea: Neither task is difficult. Each one alone is close to as fast as a person can press a button. Put them together and both get slower, which means something they both need is in short supply and cannot simply be doubled up. The more interesting part is that the two costs are rarely equal. Whichever task you treated as the one to get right first tends to hold most of its speed, and the other one absorbs the delay. That is why "can you do two things at once" is the wrong question. You can. The question is what it costs, and which of the two ends up paying.
One person, a short block of each, in a browser, and the three blocks always ran in the same order, so any practice effect across the session is mixed into the dual-task cost and would make it look smaller than it is. A proper design would counterbalance the order across people. Browser timing adds noise of its own. A cost near zero, or a negative one, is a result about the block rather than about you. And nothing here says anything about driving, or phones, or anyone's ability to multitask: those involve tasks that are far less trivial than these two, over far longer periods.
The longer version adds a difficulty manipulation, a response-overlap manipulation that separates a bottleneck at the response from one earlier, and a full stage track with block-by-block records. It is at Dual-Task and Limited Capacity Laboratory in the main collection.