Memory & Cognition1978, Vol. 6 [4) 342-353Environmental context and human memorySTEVEN M. SMITH and ARTHUR GLENBERGUniversi(v of Wisconsin, Madison, Wisconsin 53706andROBERT A. BJORKUniversity of California, Los Angeles, California 90024Five experiments examined the effects of environmental context on recall and recognition.In Experiment 1, variability of input environments produced higher free recall performancethan unchanged input environments. Experiment 2 showed improvements in cued recall whenstorage and test contexts matched, using a paradigm that unconfounded the variables ofcontext mismatching and context change. In Experiment 3, recall of categories and recall ofwords within a category were better for same-context than different-context recall. In Experi-ment 4, subjects given identical input conditions showed strong effects of environmentalcontext when given a free recall test, yet showed no main effects of context on a recognitiontest. The absence of an environmental context effect on recognition was replicated in Experi-ment 5, using a cued recognition task to control the semantic encodings of test words. Inthe discussion of these experiments, environmental context is compared with other types ofcontext, and an attempt is made to identify the memory processes influenced by environ-mental context.That one’s ability to retrieve (or recognize) an itemis heavily influenced by the relation between thatitem’s storage and retrieval contexts is indisputable."Context," however, is a kind of conceptual garbagecan that denotes a great variety of intrinsic or extrinsiccharacteristics of the presentation or test of an item.The particular concern of this article is whether thegeneral environmental context in which an item ispresented or tested can be conceptualized to functionin the same fashion as contextual factors that havemore obvious influences on the encoding of an item,such as the nature of surrounding items (Gartman &Johnson, 1972). A second concern of this article is toidentify specific cognitive processes affected by theenvironmental context.Context has been named as the agent responsiblefor a number of phenomena. It may induce oneThe first two experiments are based upon a thesis submittedby the first author to the University of Michigan in partialfulfillment of the requirements for an undergraduate honorsdegree. The last four experiments are based upon a thesissubmitted by the first author to the University of Wisconsinin partial fulfillment of the requirements of the Masters ofScience degree. The authors are indebted to Elizabeth Bjork,who helped with the design of Experiment 1, and to WilliamEpstein for helping to clarify certain theoretical issues.Thanks are also due to David Grant and Marian Schwartz forparticipating on the first author’s masters oral examinationcommittee. This research was partially supported by theUnited States Public Health Service Grant 1-R01-MH26643-01to A.M. Glenberg. Requests for reprints should be sent toStevenM. Smith, Department of Psychology, University ofWisconsin, Madison, Wisconsin 53706.encoding or another, it may be an intrinsic part of amemory trace, it may be used in matching and decisionprocesses for recognition memory, it may provide aretrieval cue for recall, and it may be used for resolutionof lexical and semantic ambiguity. The type of contextbeing referred to, however, is not always specified,and it is not empirically clear that all types of contextplay identical roles with respect to these mentalpheno~nena. Bower states that context is made upof "background external and interoceptive stimulationprevailing during presentation of the phasic experimentalstimuli ... [plus the subject’s] mental set" (1972,p. 93). Anderson and Bower add to the definition"physical characteristics of an item’s presentation,implicit associations to the items, and some cognitiveelements representing the list in question" (1974,p. 409). To attribute an identical formal effect to allsuch aspects of an item’s context may or may notturn out to be a valid speculative leap, but there islittle in the way of empirical evidence that wouldargue for or against such a notion.Context has been referred to and examined in avariety of forms. Carr (1917, 1925). one of the originalinvestigators of context, found that changes in mazeorientation and illumination disrupted the maze-runningperformance of rats. Burri (1931) studied context inthe form of audience vs. no audience; Dulsky (1935)examined effects of the background color of responsewords; and Reed (1931) studied changes of sensorymode and posture. Gartman and Johnson (1972)investigated contextual effects of nearby list words;Epstein (1961) examined syntactic context; and Eich,342ENVIRONMENTAL CONTEXT AND MEMORY343Weingartner, Stillman, and Gillin (1975) studied contextas a state of mind associated with the presence orabsence of a drug. It is not obvious that one shouldexpect all of these "contexts" to play the samefunctional role in human memory.The present set of experiments deals with thephysical aspects of the experimental situation, referredto in this paper as environmental context. Elementsof the environmental context include such things asthe time of day, the building and room in which to-be-remembered items are presented, the appearance of theexperimenter, and the sensory mode of the stimuli.Past studies of environmental context have found thatrecall and relearning performance improves if the subjectis tested in his original learning environment, ratherthan a changed one (Reed, 1931; Smith & Guthrie,1924), and that retroactive interference (RI) is reducedif interpolated learning occurs in an environmentdifferent from that of original learning (Bilodeau &Schlosberg, 1951; Greenspoon & Ranyard, 1957; Strand,1970).The interpretation of these environmental contexteffects has generally been that the learned materialis somehow connected or associated with an internalrepresentation of the physical environment. Strand(1970), however, asserted that environmental contextwas being confused with what she termed "psychologicaldisruption." Strand pointed out that subjects in theBilodeau and Schlosberg (1951) investigation weredisrupted in the changed-context condition, but notin the same-context condition. This disruption, shestated, enabled subjects to more easily distinguishbetween the two sets of words, thus causing lessinterference. To test this hypothesis, she added anothercondition to the Bilodeau and