"This means that if Task 1 processing accesses the bottleneck stage, the retrieval of semantic Task 2 memory information is delayed until Task 1 bottleneck stage processing is completed. In other words, in a dual-task context access to high level representation in memory is affected by additional task performance.
The fact that the activation of semantic memory content is delayed in high load conditions of a dual-task situation is not only of theoretical value but may also bear some important practical implications." Even though in different studies "the observed delay of accessing semantic memory is within the range of milliseconds, such delays may accumulate over time in situations of continuous multi-tasking.
Eventually, such accumulated delays may weaken or even completely interrupt performance."
Parallel memory retrieval in dual-task situations?
Fischer, R.