Small neurons, big concepts

Concept cells have a selective response to a certain concept. A ‘Shrek-concept cell’ would for example increase its firing rate to pictures of Shrek and to its spoken or written name. To find out how single neurons contribute to conceptual representations during reading we performed reading tasks with epilepsy patients that were implanted with depth electrodes, enabling us to simultaneously record spiking activity from single units from the hippocampus (Chapter Ⅱ). The reading task contained sentences with nouns that referred to the preferred concepts of cells that we were recording from, plus pronouns referring to those concepts: e.g. “Shrek and Doris walked into a bar. He sat at the table.”. In this example, the noun ‘Shrek’ refers to the preferred concept of the ‘Shrek-concept cell’ and the pronoun ‘he’ refers to Shrek. By comparing the neuronal response to this sentence to sentences where the pronoun referred to other nouns, we were able to investigate how cells facilitate sentence comprehension by activating conceptual representations during reading. We showed for the first time that concept cells can be reactivated by pronouns when those pronouns refer to their preferred concept. When you are searching for your keys to go out the door for grocery shopping, you use your memory of what the keys look like to guide the visual search while simultaneously memorizing your grocery list for when you arrive at the supermarket. Both memory items are held in working memory, but on how they are represented and how the brain distinguishes between them remains an intense debate. In Chapter Ⅲ we studied the role of single units from the medial temporal lobe in working memory and how they encode prioritized and deprioritized memory items by recording from epilepsy patients while they perform a dual-task paradigm. This study reveals that neurons in the medial temporal lobe represent both prioritized and deprioritized items in working memory by behaviorally relevant persistent firing patterns. Visual stimuli make up a significant proportion of all incoming information in humans. However, the visual system can be tricked. A briefly presented visual object can be rendered invisible when it is followed by the presentation of a mask with closely adjacent contours. This is called metacontrast masking. Researching such visual illusions can help us understand how the brain processes visual input. What many metacontrast masking paradigms have in common is that when the mask appears, it gains ownership of the borders of the previously presented target object. In Chapter Ⅳ we designed new metacontrast masking paradigms with masks that either gain ownership of the borders of the target or not. We hypothesized that perception is impaired when metacontrast masking limits the amount of time for which a target object owns its borders. For each experiment, we found that the masking effect was larger when the mask took over the borders of the target object than when border-ownership was not affected by the mask.