Visible Bottleneck I

Q: OK, I tried it. What does Visible Bottleneck I show?

· Presumably you found it impossible to keep both bars from rising up when you tried to perform both tasks at the same time. So does everyone else as far as we know. This illustrates that one cannot normally carry out two tasks completely independently when each of them requires a choice of response. When we try to do so, substantial delays occur in one or both tasks. This is true even when neither task is anything that would be described as mentally challenging.

Q: How does the demo work?

· Each bar rises at a pre-set rate. A correct response in the relevant task lowers the bar by a certain amount. An incorrect response raises it. Thus, to keep it from rising to the top, you must respond both quickly and accurately. The parameters have been adjusted so in the single-task situation most people can keep the bar from rising, or even drive it down to the floor. In the dual-task situation, the parameters stay the same as in the single-task situations; the only difference is that you need to do both tasks simultaneously. If you could do both tasks in parallel at full speed, you should experience no great difficulty. (To verify that the difficulty of the two tasks is unchanged, try having someone sit down with you at your computer and do one of the tasks while you do the other. That’s a lot easier, isn’t it?)

Q: What causes this interference?

· Much research in this area argues that one particular mental operation is almost invariably carried out sequentially in tasks like this: the planning of responses. The same is true of certain types of decision operations and memory retrievals. On the other hand, the brain seems capable of perceiving stimuli while it is choosing a response, and actually producing motor responses in one task can overlap with the choice of a response in another.

Q: Where can I read more about this phenomenon?

· This article provides an overview.

zQ: Are all tasks requiring a choice of responses subject to this sort of processing bottleneck?

· Tasks that involve extremely “natural” mappings between stimuli and responses appear not to be. For example, repeating words aloud as you hear them is a task most people can carry out in parallel with other tasks (McLeod & Posner, 1984). The same is true of moving your eye to look at a spot (Pashler, Carrier & Hoffman, 1993). There may be others. Pressing a button depending on the spatial location of a disk (as in Visible Bottleneck II) may also bypass this bottleneck.

Q: What is the neural basis for this bottleneck?

· No one really knows, although some people have recently offered speculations, e.g. R. Marois & Colleagues.

Q: Are you saying people can only do one thing at a time?

· Not at all. If one of the tasks does not involve a choice of responses (e.g., if it merely involves repetitive rhythmic tapping, or requires perceiving and identifying stimuli without the need to decide on responses), interference is often reduced or even absent (subsequent demos on this site will illustrate this point). Laboratory experiments in which response times are analyzed in detail have lent considerable support to the idea of a “central bottleneck” in response planning and indicated that other operations are often processed in parallel between the two tasks (for recent reviews, see H. Pashler, The Psychology of Attention, 1998, MIT Press; P. Jolicoeur, Journal of Experimental Psychology: Human perception and Performance, 1999, 25, 596-616).

Q: When was a bottleneck in response planning first suspected?

· A slowing when people must respond to two stimuli presented in rapid succession was first observed by Telford in 1931. Several psychologists in the UK, including Margaret Vince and Kenneth Craik made related observations in the 1940s. Alan Welford was the first to specifically claim that the brain is subject to a single-channel bottleneck arising in the selection of responses.

Q: Does this Visible Bottleneck I demo have any scientific significance or is it just a demonstration?

· It may have some. In most laboratory studies of dual-task performance, the subject performs two discrete tasks in rapid succession (the so-called “psychological refractory period” experiment). A few psychologists have argued that dual-task slowing there may reflect voluntary postponement of processing, not a basic performance limit of the brain. They propose that subjects postpone selecting responses when presented with a pair of discrete tasks in order to make sure they do not respond in a reverse order. Whether or not that is plausible in the laboratory tasks, with the Visual Bottleneck applet there is a strong incentive for the subject to ignore response order completely, and process the tasks independently if that is possible. After all, if you could perform the two tasks independently, you would keep the bars from rising. This does not seem to happen.

Also, other psychologists have suggested that the slowing found in the discrete laboratory experiments just described might reflect the abrupt nature of the stimuli used in those tasks. Perhaps the brain can process two tasks at the same time, they argue, but one task would need to be performed for a few seconds before this capability would emerge. The present demo seems to indicate that this is not a critical factor (if you try getting one task going and then adding in the other task you will find that it doesn’t help much). Thus, this simple applet has some relevance to current controversies in the field of human attention research.

Q: Has this rising-bar task ever been studied in the laboratory?

· Not so far as we know. Related studies were performed by Kalsbeek and colleagues (1967) and Gladstones and colleagues (1989), however. They didn’t have bars, but they did examine the rate of performance of two tasks. Both groups interpreted their results as favoring the idea of a central bottleneck.

Q: What would happen if one practiced this task for a long time? Would it become “automatic” so I could keep both bars down?

· One thing that would be certain to happen is that you would get much faster even at the single-task performance. If the parameters were kept as they are, you would find it easier and easier to drive the bar down to the floor in the single task situation. The interesting question is what would happen if the rate at which the bars move up were increased as you got better, keeping the single-task situation equally challenging. What would happen then in the dual-task situation? Would performance fail catastrophically? We don’t know yet, but we hope to find out.

Q: Does this kind of research have any practical implications?

· The fact that there are severe limitations in our ability to perform certain mental operations simultaneously, even when the tasks appear simple and don’t involve the use of the same body parts, has obvious implications for issues like the safety of driving while using cellphones. More generally, a better understanding of dual-task performance should be helpful in interface design for any activity where rapid performance can be important, such as in aviation as well as the use of automobiles. Distraction appears to be a factor in many accidents, so a better understanding of attention limits should be useful.

Q: I had no trouble keeping both bars down!

· Really? Please contact us.

Leave a Reply

Your email address will not be published.