A key idea of the associative network is that subthreshold activation can accumulate, so that insufficient activation received from one source can add to the insufficient activation received from another sourse. Direct evidence come from the lexical-decision task. In this task, research participants are shown a series of letter sequences on a computer screen. Some of the sequences spell words; other sequences are letter string that aren't words. The participants' task is to hit a "yes" button if the sequence spells a word. They perform this task by "looking up" these letter strings in their "mental dictionary", and they base their response on whether they find the string in the dictionary or not.
Meyer and Schvaneveldt presented their participants with pairs of letter strings, and participants had to respond "yes" if both strings were words. If both strings were words, sometimes the words were semantically related. Consider a trial with "bread, butter". Participants first need to "look up" the word "bread" in memory. They'll search for, and presumably activate, the relevant nodes. Then they're ready for the second word in the pair. The BREAD node in memory has now been activated. This will, trigger a spread of activation outward from this node, bringing activation to other, nearby nodes. These nearby nodes will surely include BUTTER, since the association between "bread" and "butter" is a strong one. This should accelerate the process of brining this node to threshold, and so it will require less time to activate.