Quantifying precision and availability of location memory in everyday pictures and some implications for picture database design

The authors investigated whether memory for object locations in pictures could be exploited to address known difficulties of designing query languages for picture databases. M. W. Lansdale's (1998) model of location memory was adapted to 4 experiments observing memory for everyday pictures. These experiments showed that location memory is quantified by 2 parameters: a probability that memory is available and a measure of its precision. Availability is determined by controlled attentional processes, whereas precision is mostly governed by picture composition beyond the viewer's control. Additionally, participants' confidence judgments were good predictors of availability but were insensitive to precision. This research suggests that databases using location memory are feasible. The implications of these findings for database design and for further research and development are discussed.     

Something Overlooked? How experts in change detection use visual saliency

How does expertise in the analysis of particular images influence the effects of visual saliency upon attention? Expert analysts of aerial photographs and untrained viewers undertook change‐detection and location memory tasks using aerial photographs with eye movements recorded throughout. Experts were more accurate in both tasks. Significant differences were also seen in the scanpaths: Untrained viewers fixated preferentially upon salient features throughout stimulus presentation whereas experts did not. However, both groups showed a strong influence of saliency in change detection and memory tasks. We interpret this apparent contradiction by: (i) assuming that the use of saliency in visual search is discretionary, and experts can use semantic information to prioritise where to fixate next; whereas, (ii) in tasks requiring spatial memory, analysis of visual saliency delivers easily acquired landmarks to reference the location of items in an image; a previously overlooked function used by expert and untrained viewers alike. Copyright © 2009 John Wiley & Sons, Ltd.     

Two spatial memories are not better than one: evidence of exclusivity in memory for object location

This paper studies the dynamics of attempting to access two spatial memories simultaneously and its implications for the accuracy of recall. Experiment 1 demonstrates in a range of conditions that two cues pointing to different experiences of the same object location produce little or no higher recall than that observed with a single cue. Experiment 2 confirms this finding in a within-subject design where both cues have previously elicited recall. Experiment 3 shows that these findings are only consistent with a model in which two representations of the same object location are mutually exclusive at both encoding and retrieval, and inconsistent with models that assume information from both representations is available. We propose that these representations quantify directionally specific judgments of location relative to specific anchor points in the stimulus; a format that precludes the parallel processing of like representations. Finally, we consider the apparent paradox of how such representations might contribute to the acquisition of spatial knowledge from multiple experiences of the same stimuli.     

When nothing is "off the record": exploring the theoretical implications of the continuous recording of cognitive process in memory

This theoretical paper explores the idea of acropetal memory; arguing that representations emanating from all stages of processing an individual stimulus can be registered in memory and potentially available for recall. I describe Lansdale's (1998) HELM model of location memory as a suitable candidate theory within which to implement and test this idea. An exemplar model, HELM-2, is presented and examined against available data. This model provides a satisfactory account of the data and an elegant solution to some interesting puzzles. It also suggests that some retrieval failure can be attributed to interference from representations of a stimulus derived from early stages of processing and that this has much broader implications for modelling encoding, forgetting, and false memory.     

Analysing uncertainty in location memory: issues in the design of spatial database systems

Future generations of databases will store increasing amounts of pictorial and other visuo‐spatial material. Among other methods, these will require retrieval techniques which exploit memory for the location of objects in pictures stored within the database. Experiments are reported in which subjects recalled the location of an object in a picture and, as an expression of their uncertainty, a range of locations within which they were sure the target fell. This is analogous to a database query method in which users specify a range of values for a location as a means of overcoming the unavoidable problem of imprecision in the expression of an analogue quantity. This paper investigates the psychological issues which allow us access to the potential value of this method in the design of pictorial databases. Although recall confidence (defined by the width of the range of responses given by the subject) is shown to be correlated with recall accuracy, the results demonstrate a trend in which subjects systematically overestimate the precision of their recall. A further experiment examines the extent to which this effect can be manipulated by instructions to the subjects. This shows that instructions aimed at reducing overconfidence produce a counterintuitive response. Subjects remain spuriously overconfident about the accuracy of responses which are in fact incorrect, but additionally they demonstrate a reduced confidence for accurate recalls. Overall, these experiments indicate that while confidence ranges can communicate useful information about the accuracy of location memory, this is less than that expected because subjects are poor at assessing the accuracy of their own recalls and are subject to a range of biases when attempting to do so. Implications for the design of retrieval systems are discussed. Copyright © 1999 John Wiley & Sons, Ltd.     

When nothing is “off the record”: Exploring the theoretical implications of the continuous recording of cognitive process in memory

This theoretical paper explores the idea of acropetal memory; arguing that representations emanating from all stages of processing an individual stimulus can be registered in memory and potentially available for recall. I describe Lansdale (1998) HELM model of location memory as a suitable candidate theory within which to implement and test this idea. An exemplar model, HELM-2, is presented and examined against available data. This model provides a satisfactory account of the data and an elegant solution to some interesting puzzles. It also suggests that some retrieval failure can be attributed to interference from representations of a stimulus derived from early stages of processing and that this has much broader implications for modelling encoding, forgetting, and false memory.     

Dilution as a model of long-term forgetting

This article presents a model of long term forgetting based on 3 ideas: (a) Memory for a stimulus can be described by a population of accessible traces; (b) probability of retrieval after a delay is predicted by the proportion of traces in this population that will be defined as correct if sampled; and (c) this population is diluted over time by null traces that, if accessed, block retrieval. Dilution is modeled as a linear function of time and outcome of accessing memories by their temporal organization. The model is applied to 5 published experiments studying forgetting in cued recall, 4 recognition experiments, and 1 using savings methods. The model specifies the different form of the retention function in each case well and provides a principled explanation for some puzzling characteristics of forgetting without recourse to mechanisms such as decay or consolidation.     

An Analysis of Errors in the Learning, Overlearning, and Forgetting of Sequences

This paper investigates the effects upon delayed recall of errors made during learning. Subjects learn a sequence of wagons in a model train over repeated presentation and test cycles. The recall of the same sequence is unexpectedly tested one week later. The results show that the errors subjects make during learning are a significant factor in predicting which elements of the sequence are forgotten. Learning is modelled by a population of discrete traces, accumulated during learning, which has two characteristics: First, many of the traces duplicate information encoded in other traces; and second, representations of the correct sequence coexist with traces containing incorrect information about the same elements of the sequence. Forgetting is modelled by the accretion of null traces that compete with correct traces at retrieval and the increasing inability with time to discriminate correct representations in memory from erroneous records. This model is shown to provide a good fit to the experimental results and suggests a strong link between the complete history of learning new material (including errors) and its subsequent likelihood of forgetting.