What are the processes involved in constructing a linear order from a set of relations between pairs of elements in the order? Subjects were presented with sets of digit pairs and instructed to form a single digit string on the basis of the order relations expressed by the pairs; for example, given 57, 19, 71, the correct response is 5719. Manipulations of presentation order revealed constructive processes of varying difficulty depending on whether a pair has zero, one, or two digits in common with the digits held in memory as an incomplete, constructed string. The least difficult processes arise when there is one common element. If the common element occurs at the end of the string in memory, construction is easier than when the new digit must be added to the beginning of the string. When a pair has no digits in common with the string in memory, constructing the correct string becomes substantially more difficult, especially when subsequent pairs having two digits in common require that a tentative order of digits held in memory be reorganized.

The time spent studying sentences that described the pairwise relations comprising linear orders to be constructed was measured. The results were consistent with the theory of constructive processes described earlier by Foos, Smith, Sabol, and Mynatt, except when the first two relations presented had no elements in common. In this situation, subjects were shown to possess two strategies for representing the order information in memory, depending on the form and content of the sentences. When the elements of the ordering were easy to rehearse and were presented in an unvarying sentence frame describing an abstract relationship, the two relations were apparently maintained by rehearsal in a single linear array. When the elements were easy to visualize and were presented in a sentence frame with a varying relation word that described a concrete relation, most subjects appeared to use a two-dimensional mental image. A modified theory in which the order of processing information in memory depends on the representational strategy was shown to account for study times under both input conditions.Previous studies have shown that the Example 1, below, as they were after order of presentation of the relationships hearing Example 2 (cf. Potts, 1972). in a linear order can have a considerable Example 1 impact on subjects' success in constructing T he farmer is taller than the doctor, the order. For example, Smith and FoosThe doctor is taller than the soldier. (1975) found that subjects were nearlyThe soldier is taller than the teacher, twice as likely to report the correct ordering Rvamtole 2 from tallest to shortest after hearing ^, ,,. . . n .. ., . , 5 UnTve'rsS'of'Tdedo 18 **" ^ ^ ^"^ * *" tive P roCeSS6S P r °POSed bv F °OS ' Smith ' Requests for reprints should be sent to Barbee T. Sabo1 ' and M V natt < 1976 ) tO aCC °"nt f ? F

The present studies were based on the hypothesis that the majority of college students have available to them the appropriate schema for understanding set inclusion relations, but that various factors influence the likelihood that the schema is used in the processing of text containing artificial inclusion relations. Although group data did not support this hypothesis. the data of individual subjects could be readily interpreted as resulting from the selection of one of a small set of representational schemata. Among the factors shown to influence schema selection were the choice of sentence frame used to present each relation. the presence or absence of real-world contextual information, and the structure (simple vs. complex) of the underlying inclusion relation. In addition, one experiment showed that the processes used in constructing a mental representation of an inclusion relation and in retrieving information from the representation are similar to those used with linear orderings.Recent studies of text processing have drawn attention to both the necessary role of inference in comprehension and the pervasive effect that inferences have on subsequent memory for texts (Bransford,

No abstract

Reading reaction time (RT) to visual stimuli was shown to vary according to the nature of simultaneous auditory stimuli. In Experiment 1, simultaneous different digits produced slower RTs than a burst of speech noise, while identicaly digits produced faster RTs. In Experiments 2 and 3 the stimuli were phoneme pairs which differed on either zero, one, or two articulatory features. Identical phonemes resulted in the fastest RTs. The RTs to non-identical phoneme pairs were not directly related to the number of differing features but were dependent upon the specific feature involved. A two-stage model was proposed to explain the results in which parallel processing of simultaneous input occurs prior to a decision-making stage.