Readers continuously receive parafoveal information about the upcoming word in addition to the foveal information about the currently fixated word. Previous research (Inhoff, Radach, Starr, & Greenberg, 2000) showed that the presence of a parafoveal word which was similar to the foveal word facilitated processing of the foveal word. In three experiments, we used the gaze-contingent boundary paradigm (Rayner, 1975) to manipulate the parafoveal information that subjects received before or while fixating a target word (e.g. news) within a sentence. Specifically a reader’s parafovea could contain a repetition of the target (news), a correct preview of the post-target word (once), an unrelated word (warm), random letters (cxmr), a nonword neighbor of the target (niws), a semantically related word (tale), or a nonword neighbor of that word (tule). Target fixation times were significantly lower in the parafoveal repetition condition than in all other conditions, suggesting that foveal processing can be facilitated by parafoveal repetition. We present a simple model framework that can account for these effects.

Retrieval practice has been shown to enhance later recall of information reviewed through testing, whereas final-test measures involving making inferences from the learned information have produced mixed results. In four experiments, we examined whether the benefits of retrieval practice could transfer to deductive inferences. Participants studied a set of related premises and then reviewed these premises either by rereading or by taking fill-in-the-blank tests. As was expected, the testing condition produced better final-test recall of the premises. However, performance on multiple-choice inference questions showed no enhancement from retrieval practice.

Recent Web apps have spurred excitement around the prospect of achieving speed reading by eliminating eye movements (i.e., with rapid serial visual presentation, or RSVP, in which words are presented briefly one at a time and sequentially). Our experiment using a novel trailing-mask paradigm contradicts these claims. Subjects read normally or while the display of text was manipulated such that each word was masked once the reader's eyes moved past it. This manipulation created a scenario similar to RSVP: The reader could read each word only once; regressions (i.e., rereadings of words), which are a natural part of the reading process, were functionally eliminated. Crucially, the inability to regress affected comprehension negatively. Furthermore, this effect was not confined to ambiguous sentences. These data suggest that regressions contribute to the ability to understand what one has read and call into question the viability of speed-reading apps that eliminate eye movements (e.g., those that use RSVP).

No abstract

The idea that people learn detailed probabilistic generative models of the environments they interact with is intuitively appealing, and has received support from recent studies of implicit knowledge acquired in daily life. The goal of this study was to see whether people efficiently induce a probability distribution based upon incidental exposure to an unknown generative process. Subjects played a ‘whack-a-mole’ game in which they attempted to click on objects appearing briefly, one at a time on the screen. Horizontal positions of the objects were generated from a bimodal distribution. After 180 plays of the game, subjects were unexpectedly asked to generate another 180 target positions of their own from the same distribution. Their responses did not even show a bimodal distribution, much less an accurate one (Experiment 1). The same was true for a pre-announced test (Experiment 2). On the other hand, a more extreme bimodality with zero density in a middle region did produce some distributional learning (Experiment 3), perhaps reflecting conscious hypothesis testing. We discuss the challenge this poses to the idea of efficient accurate distributional learning.