We took special care in choosing a reference montage for our analysis, as it has been shown that using a common reference can sometimes lead to specious see more results when investigating properties of the phase (Schiff, 2005). The decision to use a bipolar montage was based on the assumptions that (1) the unwanted reference signal was recorded equally by each electrode in the bipolar pair and (2) each electrode in the pair measured complementary components of the same phenomenon (Zaveri et al., 2006). Assumption (1) is satisfied due to the physical setup of our recording device, and assumption (2) is valid due to the extremely localized measurements of the microwire electrodes.

If anything, we would be concerned that two adjacent microwires were placed so close together that they give exactly the same measurements and the bipolar pair would thus be useless. The relevant local information would be thrown away with the unwanted reference signal. We can see from the results presented here that this is not always the case, although it may account for the variability of results from each microwire bundle. In general, the use of a bipolar montage MI-773 nmr is the most conservative choice we can make; it guarantees that the signals used in our analysis are localized to a specific brain region, but it may also reduce the strength of the results due to the loss

of relevant behavioral information. Patients were presented with a 4 × 4 grid of face-down cards on a laptop computer screen and were told that there were eight pairs of matching cards (Figure 1A). When they used the mouse to click on a face-down card, the card flipped over and an image appeared. The goal was then to click on the matching card hidden among the other face-down images. After clicking on a pair of cards, matched pairs remained visible, while unmatched images flipped over again after approximately 1 s in order to be matched on a later turn. When all 16 cards were matched, a new puzzle was generated with randomly chosen images and locations. The game contained eight categories of images (e.g., faces, teddy bears, giraffes, watermelons, ice cream, shoes,

globes, and waterfalls), with six unique images in each category. Each subject completed two sets of ten puzzles. A set contained however 80 correct trials and 87.9 ± 20.1 incorrect trials, depending on how efficiently the patient completed the task. The experiment was run using the Psychophysics Toolbox in MATLAB. In our analysis, we draw a distinction between the mouse click on the first image of each pair (“first click”) and the next click on its potential match (“second click”). For the six subjects, the average time between the first and second click ranged from 1.2–2.1 s for the first set of ten puzzles, and it ranged from 1.0–1.7 s for the second set of puzzles. The average time between all clicks ranged from 1.7–2.3 s and 1.5–1.