On the Hierarchical Structure of the Original Posner Task

Abstract

In 1967, a task for analysis of the depth of processing in simple classification was published (Posner & Mitchell, 1967, p. 392). This task has become known as the Posner task for long-term memory retrieval of information. In the original experiment, pairs of items (letters, nonsense forms, digits) were presented to the participants who were supposed to respond as fast as possible by indicating either or different. In completing the trials of this task the participants had to perform according to different rules that constituted three experimental conditions. (1) Under the instruction they had to check whether the two items presented on a screen showed physical identity (e.g., A A). (2) The instruction required the participants to compare the two items with respect to their semantic meaning. More precisely, either the same name (e.g., a A) or different names (e.g., a B) applied to both items. (3) Finally, there was the rule instruction that utilized the possibility to classify letters as vowels or consonants, i.e., the participants had to decide whether both items belonged to the same group. The depth-of-processing idea guiding Posner and Mitchell's work suggested that there are different levels of information processing. While the first level demanded mere processing of feature information in the sense of the instruction, the second level necessitated access to stored information, as is demanded by the instruction. Finally, the rule instruction additionally required the recovery of rules and their application.To a considerable extent, Posner and Mitchell's (1967) depth-of-processing idea is in line with another, even more influential notion in cognitive research, namely Craik and Lockhard's levels-of-processing model that was first published in 1972. According to this model, the accessibility of information stored in memory is considered a function of the depth of mental processing when storing information in memory. 'Deep' mental processing produces memory traces that are more elaborated, longer lasting, and stronger than those formed by 'shallow' mental processing. Furthermore, deep and shallow mental processing are proposed to represent two anchor points of a continuum. Hence, deep processing involves the identification and elaboration of semantic contents, linking contents and learnt items, and creating links for embedding new information in an already available web of information. In contrast, shallow processing is restricted to establishing links between one or a few obvious and easily accessible features of a learnt item and corresponding conceptual information in the web. As a consequence, information stored by deep mental processing can be accessed and retrieved in many ways, whereas only one or very few routes to access information exist for information stored by shallow processing. Finally, there are consequences with regard to the long-term availability of learnt items stored in long-term memory: the more links lead to a learnt item, the smaller the probability of losing access to this item in the long run.In the years following the publication of the Posner task (Posner, Boies, Eichelmann, & Taylor, 1969; Posner & Mitchell, 1967), the research interest concentrated on the physical and name identity conditions whereas the rule-identity condition was neglected. It was Hunt (1978) who especially highlighted the importance of the qualitative difference between the mental processes underlying performance elicited by the physical- and name-identity instructions. According to Hunt, the physical-identity instruction stimulates processes that are mechanistic and occur unconsciously. These processes are to be seen as a kind of background activity that occurs in combination with other kinds of cognitive processes. In contrast, the name-identity instruction is assumed to be associated with memory processes that conduct a search in the web of information also referred to as knowledge base or long-term memory. …