What happens in our mind before we produce a word? It is known that naming a picture of a dog is faster than naming a picture of a couch. One possibility why this happens is because a picture of a dog will almost always be named as “dog”, whereas a picture of a couch may elicit the name “sofa” or even “settee”, indicating lower picture-name agreement. Picture name agreement is a measure of the proportion of speakers who independently produce a picture’s modal name when asked to name it. This measure of lexical availability for pictures is associated with robust effects in word production, which have been assumed to index the competition between lexical representations for selection. On this account, “couch” and “sofa” actively race for selection, delaying production speed, until the best option is ultimately chosen. But is such competition warranted? The research reported in the current thesis examined picture name agreement as a measure of lexical co-activation in word production in a bid to clarify whether selecting words for speech requires such an active competition between representations. By measuring speakers’ word choices, naming latencies and electrophysiological activity as they named pictures with high and low name agreement in a variety of simple tasks, I was able to show that these effects point to the co-activation of linguistic representations in their minds, but also index speakers’ unique idiosyncratic preferences for specific words. Overall, variations in picture name agreement fail to provide strong support for a competitive account of lexical selection, but instead favor a view in which co-activation of words in the mental lexicon appears to be effortful, but eventually leads to the successful production of the best candidate for each individual speaker.