As part of my PhD within the field of Human-Computer Interaction (HCI), one of the things I am trying to support the user in is multitasking. This might be relevant for you since interruptions and multitasking are highly interlinked. I'll refer you to some of the key papers I encountered during my background research, including some on the effects of interruptions.
You are asking "how long it usually takes a person to refocus". In order to answer that question fully you'd need to define 'refocus' better, but there are some papers on some effects of interruptions.
Multitasking and activities
One of the main referenced papers on multitasking within HCI is “Constant, Constant, Multi-tasking Craziness”: Managing Multiple Working Spheres by Victor M. González and Gloria Mark, 2004. They introduced the concept of 'working spheres' representing higher-level contexts, which "people spend about 12 minutes in [...] before they switch to
another". I assume you are interested in the effects on these higher-level context switches, rather than low level tasks as part of these contexts. This is an important first difference to keep in mind when reading up on multitasking/interruptions. Following up on citations of this paper will lead you to a lot of papers which might interest you.
Differences in interruptions
Not all interruptions are alike and not all interruptions are even disruptive. Monk et al. (2008) note "the time to resume task goals after an interruption varied depending on the duration and cognitive demand of interruptions".
Some characteristics that have been observed to affect primary task performance:
- Cognitive demand
- Similarity to primary task
- Relatedness to primary task
State when interrupted
- Control over interruption onset
- Availability of primary task retrieval cues
- Cognitive load (Bailey, B. P., & Iqbal, S. T., 2008)
For references to the relevant papers and a lengthy discussion see Monk et al., 2008.
Effects of interruptions
Studies concluded that people performed post-interruption tasks more slowly compared to pre-interruption performance. They also found that people made more errors in post-interruption performance. (Kreifeldt, J. G., & McCarthy, M. E., 1981) (Gillie, T., & Broadbent, D., 1989)
However, as Monk et al. also describe in their review there have been contradictory findings. For example, in a more recent paper by Gloria Mark et al., 2008 "people completed interrupted tasks in less time with no difference in quality". They suggest people compensate for interruptions by working faster, at a price: experiencing more stress, higher frustration, time pressure and effort.
More relevant to your question, although described in a lot more detail by Monk et al. (2008) the graphs near the end of the paper show a resumption time of about 1250-1700 ms depending on the interruption duration (5-60s). These, however, seem to reflect much smaller (low-level) context switches.
My intuition (and some contradicting findings) tells me it's hard to generalize these results. I hope by providing this overview you have a broader picture of the scope of interruptions. I suggest you start by looking into Monk C. A. et al.'s paper (2008). It provides a nice overview and reports on the resumption time results mentioned earlier, relating them to memory models which might interest you. If you follow up on citations and find something interesting, please keep us up to date! :)
Again, it is still important to be concrete about what 'resumption times' are. I don't feel like there is one official definition, and rather think it's whatever researchers decide to measure in their methodology. I've done some experiments myself measuring resumption time as the time needed to re-establish the working context needed to execute the task being switched to, using the Windows 7 window manager (Jeuris & Bardram, 2016). Although this sometimes only took people around 2 s, others were totally overwhelmed with the many resources open due to multitasking and it took them over a minute.
González, V. M., & Mark, G. (2004, April). Constant, constant, multi-tasking craziness: managing multiple working spheres. In Proceedings of the SIGCHI conference on Human factors in computing systems (pp. 113-120). ACM.
Monk, C. A., Trafton, J. G., & Boehm-Davis, D. A. (2008). The effect of interruption duration and demand on resuming suspended goals. Journal of Experimental Psychology: Applied, 14(4), 299.
Gillie, T., & Broadbent, D. (1989). What makes interruptions disruptive? A study of length, similarity, and complexity. Psychological Research, 50(4), 243-250.
Kreifeldt, J. G., & McCarthy, M. E. (1981). Interruption as a test of the user-computer interface.
Bailey, B. P., & Iqbal, S. T. (2008). Understanding changes in mental workload during execution of goal-directed tasks and its application for interruption management. ACM Transactions on Computer-Human Interaction (TOCHI), 14(4), 21.
Mark, G., Gudith, D., & Klocke, U. (2008, April). The cost of interrupted work: more speed and stress. In Proceedings of the SIGCHI conference on Human Factors in Computing Systems (pp. 107-110). ACM.
Jeuris, S., & Bardram, J. E. (2016). Dedicated workspaces: Faster resumption times and reduced cognitive load in sequential multitasking. Computers in Human Behavior, 62, 404-414.