Before we look at the data, let’s consider the design of the ERP CORE N400 experiment. As illustrated in Figure 2.1.A, the experiment involved a sequence of trials, each of which consisted of a prime word followed by a target word. The participants’ task was to press one of two buttons on each trial to indicate whether the target was semantically related or semantically unrelated to the preceding prime word. For example, they would press the related button for CHAIR preceded by TABLE and the unrelated button for SPIDER preceded by RAKE. Additional methodological details can be found in Kappenman et al. (2021).
Hundreds of previous studies have shown that a word will elicit a larger N400 component if it is unrelated to a previously established semantic context than if it is related to that context (Kutas, 1997; Swaab et al., 2012), so we expected to see a larger N400 on unrelated trials than on related trials. Figure 2.1.B shows the ERP waveforms, recorded at the CPz electrode site (where the N400 is typically largest) and averaged over all 40 of the original participants (a grand average). In these waveforms, the N400 is a negative-going wave for the unrelated targets that is present from approximately 200-600 ms and is added onto the positive voltage that is ordinarily present during this time period.
It is often useful to subtract away everything that is in common to two conditions and focus on the difference in brain activity between conditions. To do this, we compute a difference wave, which is simply the difference in voltage between the two conditions at each time point. Figure 2.1.C shows the unrelated-minus-related difference wave for our N400 experiment. It allows us to see the brain’s differential processing of unrelated versus related words. Because the difference deviates from zero at approximately 200 ms, we can conclude that the brain has determined whether the target is related or unrelated to the prime by this time.