Jeff Miller

Divided attention: Evidence for coactivation with redundant signals

Abstract Many models of divided attention assume that signals presented on different channels produce separate activations, any one of which can initiate a response. According to these models, detection responses are especially fast when signals are presented on two channels at the same time because the system can detect a signal in either of two ways. Such models predict a testable relation among reaction time distributions for conditions in which a single signal is presented as compared with a condition in which two signals are presented, and this prediction is tested in two tasks. A bimodal detection task required a simple speeded response to either a visual or an auditory signal. A letter search task required a choice response depending on whether or not a two-letter display included the signal letter “A.” Data from both tasks are inconsistent with the prediction. When two signals are presented, responses are faster than separate-activation models can explain. The results favor “coactivation” models, in which signals presented on different channels contribute to a common pool of activation that initiates a response.

Jackknife-based method for measuring LRP onset latency differences

A new method based on jackknifing is presented for measuring the difference between two conditions in the onset latencies of the lateralized readiness potential (LRP). The method can be used with both stimulus- and response-locked LRPs, and simulations indicate that it provides accurate estimates of onset latency differences in many common experimental conditions.

Using the jackknife‐based scoring method for measuring LRP onset effects in factorial designs

Miller, Patterson, and Ulrich (1998) introduced a jackknife-based method for measuring the differences between two conditions in the onset latencies of the lateralized readiness potential (LRP). The present paper generalizes such jackknife-based methods to factorial experiments with any combination of within- and between subjects factors. Specifically, we introduce a subsample scoring method to assess potential main and interaction effects on LRP onsets within conventional yet slightly adjusted analyses of variance (ANOVAs) and post hoc comparison procedures.

Effects of truncation on reaction time analysis

Many reaction time (RT) researchers truncate their data sets, excluding as spurious all RTs falling outside a prespecified range. Such truncation can introduce bias because extreme but valid RTs may be excluded. This article examines biasing effects of truncation under various assumptions about the underlying distributions of valid and spurious RTs. For the mean, median, standard deviation, and skewness of RT, truncation bias is larger than some often-studied experimental effects. Truncation can also seriously distort linear relations between RT and an independent variable, additive RT patterns in factorial designs, and hazard functions, but it has little effect on statistical power. The authors report a promising maximum likelihood procedure for estimating properties of an untruncated distribution from a truncated sample and present in an appendix a set of procedures to control for truncation biases when testing hypotheses.

Timecourse of coactivation in bimodal divided attention

Reaction time distributions were obtained from practiced subjects in a go/no-go detection task with attention divided across the visual and auditory modalities. Redundant signals were sometimes presented asynchronously on the two modalities, with the time between signals varying from 0 to 167 msec. An extension of the inequality derived by Miller (1982) was used to test between separate-decisions models, in which the response is initiated solely by whichever signal is detected first, and coactivation models, in which both signals contribute to the activation of a single response. As in previous studies with bimodal detection tasks, the results contradicted separate-decisions models and favored coactivation models. The largest violations of separate-decisions models were observed when the visual signal was presented 67–100 msec before the auditory signal. A new inequality was also derived to discriminate between two classes of coactivation models that differ about whether responses are generated by processes combining activation across time as well as across signals. Violations of this inequality ruled out exponential coactivation models, in which activation processes are sensitive only to the instantaneous properties of the signal(s). Instead, the results require an accumulation model of coactivation, in which both signals provide input to a process that accumulates activation over a considerable period of time, even if signal conditions change during that time.

The flanker compatibility effect as a function of visual angle, attentional focus, visual transients, and perceptual load: A search for boundary conditions.

When subjects must respond to a relevant center letter and ignore irrelevant flanking letters, the identities of the flankers produce a response compatibility effect, indicating that they are processed semantically at least to some extent. Because this effect decreases as the separation between target and flankers increases, the effect appears to result from imperfect early selection (attenuation). In the present experiments, several features of the focused attention paradigm were examined, in order to determine whether they might produce the flanker compatibility effect by interfering with the operation of an early selective mechanism. Specifically, the effect might be produced because the paradigm requires subjects to (1) attend-exclusively stimuli within a very small visual angle, (2) maintain a long-term attentional focus on a constant display location, (3) focus attention on an empty display location, (4) exclude onset-transient flankers from semantic processing, or (5) ignore some of the few stimuli in an impoverished visual field. The results indicate that none of these task features is required for semantic processing of unattended stimuli to occur. In fact, visual angle is the only one of the task features that clearly has a strong influence on the size of the flanker compatibility effect. The invariance of the flanker compatibility effect across these conditions suggests that the mechanism for early selection rarely, if ever, completely excludes unattended stimuli from semantic analysis. In addition, it shows that selective mechanisms are relatively insensitive to several factors that might be expected to influence them, thereby supporting the view that spatial separation has a special status for visual selective attention.

Discrete and continuous models of human information processing: theoretical distinctions and empirical results

Abstract Information-processing models of human performance are often categorized as discrete or continuous, with continuous models currently in vogue. This paper presents a three-part analysis of discrete and continuous models. Part 1 considers the definitions of the terms discrete and continuous, and concludes that a dichotomous categorization of models is a serious oversimplification for three reasons. First, discreteness and continuity are not themselves dichotomous properties, but rather two extremes between which there can be intermediate cases. Second, there are many different senses in which information-processing models can be described as discrete or continuous (or somewhere in between), and the same model may be relatively discrete in some senses and relatively continuous in others. Third, any complete model of an information-processing task assumes several different processing stages, of which some could be relatively discrete (in any given sense) and others relatively continuous. Part 2 reviews evidence commonly cited in support of continuous models. Many of the findings can be reconciled with fully discrete models, and the rest require continuity only in certain very limited senses. In particular, there is no compelling evidence against the discrete stage assumption of the Additive Factor Method (AFM) of Sternberg (1969a,b). Part 3 reviews recent evidence supporting discrete models, some of which specifically supports the discrete stage assumption. In the light of this evidence and the shortcomings of the evidence reviewed in Part 2, it is concluded that it is premature to abandon discrete models in favor of continuous ones.

Queuing or Sharing? A Critical Evaluation of the Single-Bottleneck Notion ☆

The model of a single central bottleneck for human information processing is critically examined. Most evidence cited in support of the model has been observed within the overlapping tasks paradigm. It is shown here that most findings obtained within that paradigm and that were used to support the model are also consistent with a simple resource model. The most prominent findings are the millisecond-for-millisecond slope at the left of the RT2-SOA curve, the high RT1-RT2 correlation, the additivity of the effects on RT2 of SOA and of the difficulty of selecting R2, and the washout of the effect of S2 discriminability on RT2 in a dual-task condition. In addition, the asymmetry of the effects of the dual-task requirement on RT1 and RT2 can be accounted for by the resource model provided that it assumes uneven allocation of resources, which is quite reasonable in view of the task asymmetry inherent in the demand characteristics of the paradigm. The same is true for two other findings that appear to support the single-bottleneck model-that in the dual-task condition, the demand of the first task affects equally RT1 and RT2 and that its effect on RT1 is the same as the corresponding effect in the single-task condition. Furthermore, the single-bottleneck model is hard to reconcile with a negative slope at the left of the RT1-SOA curve or a positive slope at the left of the IRI-SOA curve, unless augmented by ancillary assumptions that are yet to be substantiated. Representative data were fit by each of the models using its optimal set of parameters. Both models achieved quite good degrees of fit. It is further argued that since the overlapping tasks paradigm is heavily biased in favor of a speedy reaction to the stimulus that appears first, it is nonoptimal for testing the central bottleneck model. Finally, the bottleneck model is examined in terms of other scientific criteria.

Electrophysiological evidence for temporal overlap among contingent mental processes

A series of studies assessed perceptual-motor transmission of stimulus information by measuring lateralization of movement-related brain potentials in a choice reaction task with no-go trials. When stimuli varied in shape and size, lateralized potentials on no-go trials suggested that easily recognized shape information was used to initiate motor preparation and that this preparation was aborted when size analysis signified that the response should be withheld. This indicates that movement preparation can begin once partial perceptual information about a stimulus becomes available, contrary to an assumption of fully discrete models of information processing. By contrast, when stimuli varied only in size, no evidence for preliminary response preparation was obtained, contrary to an assumption of fully continuous models but consistent with asynchronous discrete coding models (Miller, 1982, 1988).

Role of outcome conflict in dual-task interference

The traditional explanation for dual-task interference is that tasks compete for scarce processing resources. Another possible explanation is that the outcome of the processing required for one task conflicts with the processing required for the other task (e.g., cross talk). To explore the contribution of outcome conflict to task interference, we manipulated the relatedness of the tasks. In Experiment 1, subjects searched concurrently for names of boys in one channel and names of cities in another channel. Responses were significantly delayed when a nontarget on one channel belonged to, or was even just related to, the category designated as the target for the other channel. No comparable effects were found when the tasks were performed in isolation. Thus, the difficulty of the individual tasks is not the only determinant of how much they will interfere when combined, and there must be substantial interactions between processes carrying out the two tasks. In Experiment 2 subjects searched one channel for specific target letters and another channel for specific target digits. The nontargets in a channel were either from the same alphanumeric category as the targets for that channel or from the opposite category (i.e., the category of the targets for the other channel). It was found that although between-category search was more efficient than within-category search in single tasks, it was less efficient in dual tasks. Thus, there appear to be significant task interactions due to the confusability emerging when the nontargets of one task belong to the same category as the targets of the concurrent task. In addition, the congruence of target presence or absence on the two channels was found to have a sizeable effect. We suggest four potential sources of outcome conflict that may contribute to dual-task interference, and we conjecture that a great deal of the residual interference might result from other sorts of outcome conflict.