Semantic representations as procedures for verification

Doubly quantified sentences can be ambiguous (Every man knows some woman) or unambiguous (Every man knows every woman). For active and passive sentences of these types, we elicited from subjects three types of judgments designed to reflect which quantifier they assigned wide scope in interpreting the sentence. There was a strong tendency for the three measures to agree, and for these agreements to fall on the surface structure subject of the sentence, independent of sentence type. The data are interpreted as showing a tendency for the first quantifier to include the second within its scope; thus for both ambiguous and unambiguous sentence types active sentences tend to be interpreted differently from their passive transforms. A semantic theory adequate to capture this phenomenon must assign sentences semantic representations specifying not only truth-conditions but also procedures for verification.     

Aversive CSs suppress lever pressing for food but not the eating of free food

Forty rats received CER acquisition, half “on the baseline” (response lever present), half “off the baseline.” During initial CER extinction, Ss received: (a) normal CER extinction (lever available), (b) free food during CSs only (no lever), (c) free food during non-CSs only (no lever), or (d) free food during both CS and non-CS periods (no lever). While normal extinction Ss were highly suppressed in the presence of the CS, all free food groups readily ate but did not differ in eating latencies. On subsequent CER extinction trials on the baseline, those Ss which received free food during the CSs were no less suppressed than other Ss. These data offer no support for Estes' (1969) reciprocal inhibition explanation of CER and punishment, nor do they support a fear interpretation of CER.     

Information extraction during visual search: A developmental progression

Five developmental experiments examine the role of context in visual search. Experiments I and II presented either lists or paragraphs that were either in meaningful context or scrambled. Results showed a significant interaction of these variables. Experiment II examined within-item context or predictability and replicated and extended the work of Brand (1971) and Ingling (1972). Experiments IV and V examined search through lists of various widths; results showed a significant interaction of Grade × Position × Width reflecting the different perceptual spans of children and adults. Although these were search tasks, appropriate comparisons between list and paragraph performance were made allowing interpretations of the results to be supportive of a model of reading in which meaningfulness, predictability, and peripheral vision play critical roles.     

Three computational versions of proportion correct for use in forced-choice experiments

Three commonly used measures of performance in forced-choice experiments, all called proportion correct, are compared as response bias, bias in the stimulus outcomes, and level of performance are varied. No position is taken on the question of which is the computational version of choice, but some of the characteristics of the various versions are examined.