Spatial and identity negative priming in audition: Evidence of feature binding in auditory spatial memory

Mayr, Susanne; Buchner, Axel; Möller, Malte; Hauke, Robert

doi:10.3758/s13414-011-0138-2

Spatial and identity negative priming in audition: Evidence of feature binding in auditory spatial memory

Published: 18 May 2011

Volume 73, pages 1710–1732, (2011)
Cite this article

Download PDF

Attention, Perception, & Psychophysics Aims and scope Submit manuscript

Spatial and identity negative priming in audition: Evidence of feature binding in auditory spatial memory

Download PDF

Susanne Mayr^1,2,
Axel Buchner¹,
Malte Möller¹ &
…
Robert Hauke¹

1449 Accesses
23 Citations
1 Altmetric
Explore all metrics

Abstract

Two experiments are reported with identical auditory stimulation in three-dimensional space but with different instructions. Participants localized a cued sound (Experiment 1) or identified a sound at a cued location (Experiment 2). A distractor sound at another location had to be ignored. The prime distractor and the probe target sound were manipulated with respect to sound identity (repeated vs. changed) and location (repeated vs. changed). The localization task revealed a symmetric pattern of partial repetition costs: Participants were impaired on trials with identity–location mismatches between the prime distractor and probe target—that is, when either the sound was repeated but not the location or vice versa. The identification task revealed an asymmetric pattern of partial repetition costs: Responding was slowed down when the prime distractor sound was repeated as the probe target, but at another location; identity changes at the same location were not impaired. Additionally, there was evidence of retrieval of incompatible prime responses in the identification task. It is concluded that feature binding of auditory prime distractor information takes place regardless of whether the task is to identify or locate a sound. Instructions determine the kind of identity–location mismatch that is detected. Identity information predominates over location information in auditory memory.

Change deafness for real spatialized environmental scenes

Article Open access 28 June 2017

The time-course of distractor processing in auditory spatial negative priming

Article 02 August 2015

Auditory enhancement of visual searches for event scenes

Article 10 January 2022

Negative priming—the slowed-down responding to previously ignored information—was originally discovered and studied in the visual modality (for reviews, see Fox, 1995; Tipper, 2001) in order to understand how human information processing deals with distracting information in the course of goal-directed behavior. However, by now, the negative priming effect has also become a well-established phenomenon in the auditory modality (Banks, Roberts, & Ciranni, 1995; Buchner & Mayr, 2004; Buchner, Zabal, & Mayr, 2003; Leboe, Mondor, & Leboe, 2006; Mayr & Buchner, 2006, 2010a; Mayr, Buchner, & Dentale, 2009a; Mondor, Leboe, & Leboe, 2005; Zabal & Buchner, 2006). By and large, auditory negative priming follows the same principles as its visual counterpart (Mayr & Buchner, 2007), which is by no means a trivial finding. In vision, attentional selection is supported by peripheral mechanisms such as head movements and shifting fixations. In contrast, auditory central selection (as measured by negative priming) has to be accomplished without peripheral support (Banks et al., 1995), which might result in either a stronger or a differently organized selection mechanism in the auditory, as compared with the visual, modality. All of the cited auditory studies focused on identity negative priming and showed that participants were slowed down in responding to a probe sound’s identity when this sound identity had to be ignored in the prime. Comparably, identity negative priming in the visual modality is found when participants are slowed down in responding to the identity of a visual object (such as a line drawing) that had to be ignored previously (e.g., Tipper, 1985). Identity negative priming, irrespective of stimulus modality, is usually explained either by inhibition of the prime distractor, which has to be overcome when the previously ignored stimulus becomes the probe target (Tipper, 1985, 2001; Tipper & Cranston, 1985), or by retrieval of prime episode information that is incompatible with probe responding (Mayr & Buchner, 2006; Neill & Valdes, 1992; Neill, Valdes, Terry, & Gorfein, 1992; Rothermund, Wentura, & De Houwer, 2005; for evidence of a very similar process of episodic retrieval in the visual search paradigm, see Huang, Holcombe, & Pashler, 2004).

The counterpart to identity negative priming is the spatial negative priming effect. In a spatial priming task, participants have to select an object on the basis of its identity or an aspect thereof (e.g., object color or shape) and have to report the location at which the selected object has been presented, while ignoring objects (and their locations) that do not conform to the selection criterion (e.g., Tipper, Brehaut, & Driver, 1990). On so-called ignored repetition trials in a standard visuospatial priming task, the attended probe target object is presented at the same position as the previously ignored prime distractor. A spatial negative priming effect is found when responses to ignored repetition trials are slowed down, in comparison with control trials in which there is no location repetition between any of the two prime and probe stimuli. In other words, spatial negative priming is the aftereffect of ignoring a task-irrelevant spatial position in favor of attending and responding to another, task-relevant position. The majority of studies concerned with spatial negative priming in the visual modality have revealed that responding to previously irrelevant locations is indeed impaired (see, e.g., Buckolz, Goldfarb, & Khan, 2004; Chao, 2009; Milliken, Tipper, Houghton, & Lupiáñez, 2000; Milliken, Tipper, & Weaver, 1994; Neill, Terry, & Valdes, 1994; Neill et al., 1992; Tipper et al., 1990; Tipper, Weaver, & Milliken, 1995).

The spatial negative priming task can be regarded as a useful tool for investigating how human information processing deals with objects appearing at to-be-ignored locations while task-relevant objects are presented at other locations. Originally, and in analogy to identity negative priming, spatial negative priming has been attributed to inhibition of ignored locations during prime processing (Tipper et al., 1990) or to retrieval of probe-incompatible prime episode information (i.e., do-not-respond tags) associated with the ignored location (Neill & Valdes, 1992). Note that both models predict a general impairment in responding to a previously ignored location, irrespective of the identity of the stimulus presented at that location.

An alternative explanation for the spatial negative priming effect has been proposed by Park and Kanwisher (1994). Instead of being the consequence of selecting against the irrelevant location in the prime, the so-called feature mismatching hypothesis asserts that the observed slowdown in ignored repetition trials is based on feature (e.g., shape or color) mismatches at the former prime distractor location that contains the target in the probe. For example, responding to a green probe target should be slowed down if a red distractor (but not a green distractor) had been presented at the same position during the prime. Note that the feature mismatching hypothesis is also based on the retrieval of prime information (otherwise, a mismatch could not be detected), but in contrast to the episodic retrieval model, it is assumed that there is only retrieval of perceptual prime details at the repeated location, not of the selective state during the prime (“ignore the prime distractor,” “respond to the prime target”). Consequently, the feature mismatching hypothesis has nothing to do with selection against an ignored location in favor of a target and, therefore, does not predict a general impairment in responding to a previously ignored location. Instead, impaired responding is predicted only when there is a feature mismatch at the repeated location. Park and Kanwisher pitted the feature mismatching hypothesis against the distractor inhibition and the episodic retrieval models in a task based on the so-called X–O paradigm. They presented displays containing four position markers with a target symbol (e.g., “O”) over one position marker and a distractor symbol (e.g., “X”) over another. Participants had to respond to the location containing the target symbol. By exchanging the target identities between the prime (target “X,” distractor “O”) and the probe (target “O,” distractor “X”), the probe target (“O”) appeared at the same location as the prime distractor (“O”) on ignored repetition trials. The feature mismatching hypothesis predicts no response slowing for this so-called match trial, because there is no prime-to-probe symbol change but a symbol match at the probe target location. The distractor inhibition model, as well as the episodic retrieval model, predicts impaired responding for match trials whenever a previously ignored location becomes the to-be-attended location. In accordance with the feature mismatching hypothesis, match trial responses were not slowed down (in fact, they were faster than responses on control trials without any location repetition).

However, considerable evidence disqualifies feature mismatching as the sole reason for visuospatial negative priming. For example, Milliken et al. (2000) showed a visuospatial negative priming effect when participants had to locate a probe target symbol “X” of a specific color, regardless of whether the prime distractor “X” and the probe target “X” were of the same or of different colors (see also Milliken et al., 1994; Tipper et al., 1995). Here, the slowdown in responding was the aftereffect of having ignored the repeated location but was not caused by a mismatch at the repeated location. This finding can be exclusively explained by assuming distractor location inhibition or the retrieval of inconsistent prime episode information attached to the previous distractor location. What is more, feature mismatching is frequently regarded to be an irrelevant or, at most, a very subordinate factor in visuospatial negative priming, which is indicated by the fact that many recent visuospatial priming studies have not even excluded or controlled feature mismatching as an explanation of their spatial priming effects (Buckolz, Avramidis, & Fitzgeorge, 2008; Buckolz et al., 2004; Fitzgeorge & Buckolz, 2008; Gibbons, 2006; Guy & Buckolz, 2007; Guy, Buckolz, & Khan, 2006; Guy, Buckolz, & Pratt, 2004). It has been assumed that a process that detects mismatching features between the current probe target and the previous prime distractor object comes into play only under very specific circumstances that have not been investigated thoroughly (Milliken et al., 1994; Tipper et al., 1995).

In contrast to visuospatial negative priming, we still know very little about auditory spatial negative priming and the underlying question of how the human cognitive system deals with locations previously occupied by irrelevant information in auditory space. There are fundamental differences between vision and audition with respect to the representation of space. Whereas the visual system is organized spatiotopically right from the retina upward, the auditory system is predominantly tonotopic. Location in auditory space is available only after being computed from interaural time and level differences, head-related transfer functions, and distance cues (although these computations may occur as early as the superior olivary complex). This difference between modalities fits with the finding that spatial cuing effects appear to be weaker and less consistent across participants (Roberts, Summerfield, & Hall, 2009) in the auditory, as compared with the visual, modality. Similarly, inhibition of return is smaller for auditory than for visual stimuli (Reuter-Lorenz, Jha, & Rosenquist, 1996). Instead, there have been found precuing effects and inhibition-of-return effects with respect to sound frequency (Mondor & Breau, 1999; Mondor, Breau, & Milliken, 1998a; Prime & Ward, 2002). These findings support the assumption that “space is not central to the formation of auditory objects” (Kubovy & Van Valkenburg, 2001, p. 99). On the basis of a great amount of experimental evidence in auditory and visual perception, Kubovy and Van Valkenberg came to this conclusion when trying to specify the indispensable attributes of objecthood in audition and vision. They argued that visual objects were formed in space and time, whereas auditory objects were based in pitch and time. Following their ideas, the pitch (i.e., identity) of a sound is a fundamental/indispensable property in auditory object formation, because pitch separation, rather than location separation, is crucial for perceiving/discriminating between multiple sounds. Given these differences in the significance of space in the visual, in contrast to the auditory, modality, it is conceivable that ignoring a location in auditory space does not result in a general impairment when the same location becomes task relevant on the next trial.

Consistent with this expectation, the only auditory spatial priming study published to date revealed evidence against a general impairment in responding to a previously ignored location (Mayr, Hauke, & Buchner, 2009b). Instead, the results supported the feature mismatching hypothesis. Participants selected a target sound on the basis of its identity while ignoring a distractor sound presented at another loudspeaker location. They reported the target’s location by an appropriate keypress. Ignored repetition trials comprised a probe target sound presented at the loudspeaker location that had been occupied by the prime distractor sound. On control trials, there was no location repetition between any of the two prime and probe sounds. The pattern of results could be explained neither by inhibition of ignored locations during the prime nor by retrieval of probe-incompatible prime episode information associated with the ignored location. Instead, the data were consistent with the feature mismatching hypothesis. Negative priming (slowed-down responses to previously ignored locations, relative to control locations) occurred only for mismatch trials—that is, for trials with a sound identity change at the repeated location (which was first occupied by the prime distractor sound and then by another probe target sound). In clear contrast, responding to the previously ignored location was not slowed down on match trials on which the same sound was presented first as a prime distractor and then as a probe target at the same location. In fact, for match trials, there was an almost significant positive priming effect.

With respect to the phenomenon of auditory spatial priming, the findings by Mayr et al. (2009b) support the feature mismatching hypothesis and stand in clear contrast to the majority of studies in the visual modality. At a more general level, however, the results may be seen as indicative of object binding in the auditory modality—that is, of the combined encoding, storing, and retrieval of sound identity and location information. Otherwise—that is, if sound identity and location information had been stored and retrieved independently—the slowed-down response times for mismatch, but not for match, trials cannot be explained. A number of recent studies (Dyson & Ishfaq, 2008; Maybery et al., 2009; Mondor, Zatorre, & Terrio, 1998b; Parmentier, Maybery, & Elsley, 2010; Zmigrod & Hommel, 2009, 2010) have presented comparable evidence showing that auditory features are mandatorily integrated into auditory object files and that these object files are stored and retrieved from memory and influence subsequent performance in a similar way as Kahneman, Treisman, and Gibbs (1992) have proposed for the visual modality (for a review, see Hommel, 2004).^{Footnote 1}

For example, Zmigrod and Hommel (2009, 2010) presented evidence indicative of spontaneous integration of auditory features into episodic object files by showing that performance in a discrimination task was better when the second of two consecutive stimuli shared all (e.g., pitch and location) or none of the features of the first, rather than when only one of the features (either pitch or location) overlapped—that is, if a so-called partial repetition occurred. Partial repetition costs are explained by a feature integration process that establishes an episodic trace containing links between the simultaneously presented features, such as pitch and location. These episodes are assumed to be retrieved as a whole when at least one of the elements is encountered again. Retrieval does not take place when there is no repetition of any of the features. Object file retrieval is unproblematic or even beneficial for subsequent responding if all features bound together are repeated. However, partial repetitions (i.e., repetitions of some but not all features) lead to reactivation of misleading and conflicting information, which slows down processing and responding.

The findings of Mayr et al. (2009b) can be seen as another demonstration of object-binding and partial repetition costs in the auditory modality. What is new, however, is the fact that this evidence of object binding refers to a completely task-irrelevant object that should have been ignored in order to process and respond to the target stimulus. In contrast, studies of object binding are usually based on presentations of only a single task-relevant stimulus that comprises more than one feature (such as pitch and loudness, as well as pitch and location in Zmigrod & Hommel, 2009). In these studies, it could be demonstrated that response-relevant, as well as response-irrelevant, features of the same object can be integrated into an episodic file. On the basis of the findings of Mayr, Hauke, and Buchner, however, the mechanism underlying object binding seems to be even more potent, since this mechanism is able to integrate stimulus features of an object entirely irrelevant for responding.

The findings by Mayr et al. (2009b) can be seen as evidence of partial repetition costs in auditory spatial priming and, at the same time, as evidence of the validity of the feature mismatching hypothesis. However, before this strong conclusion can be accepted as fully convincing, it seems necessary to substantiate the empirical basis. Therefore, the first purpose of Experiment 1 was to replicate conceptually the Mayr, Hauke, and Buchner findings. This previous experiment had been conceptualized as the auditory equivalent of the visual X–O paradigm, which allows for only a limited analysis of sound–identity mismatches, because there are only two sounds jointly presented on every trial throughout the experiment. This allows for a manipulation of feature matches/mismatches at the repeated (first ignored, then attended) location. This kind of mismatch directly corresponds to the mismatch concept given by Park and Kanwisher (1994). However, there is a second type of identity–location mismatch conceivable that was not implemented in the study by Mayr, Hauke, and Buchner: A mismatch can also arise when the same sound is presented as prime distractor and as probe target but its location changes. If the object-binding framework were a valid model of the mechanisms underlying auditory spatial negative priming, slowed-down responding should also be found for this second type of mismatch. We thus extended the stimulus set size from two to four environmental sounds in Experiment 1. A set of this size allows for an orthogonal manipulation of sound identity repetition and sound location repetition with respect to the prime distractor and the probe target. This allowed us to implement both mismatch types necessary to test the object-binding framework, identity changes at a repeated location and location changes of a repeated identity.

To anticipate, Experiment 1 revealed evidence compatible with the feature mismatching hypothesis in particular and the object-binding framework in general. Both types of identity–location mismatches led to slowed-down responses. In Experiment 2, we changed the experimental task from a spatial to an identity priming task but kept the auditory stimulation equivalent to that in Experiment 1. With Experiment 2, we wanted to test whether evidence of object binding would also be found in a task that probed participants to respond to the identity of sounds (such a finding would stand in clear opposition to the previous identity negative priming literature). An additional question was whether partial repetition costs would be similar to those found in the spatial task in Experiment 1.

Experiment 1

For prime displays in Experiment 1, participants had to locate a target sound that could appear at one of four loudspeakers arranged around the participant, while ignoring another sound at one of the other three locations. In the probe display, the task was the same, but the probe target sound could appear either at the former prime distractor location or at a location that had not been occupied in the prime. Independently of the location repetition manipulation, the probe target could be either the same sound as that of the prime distractor or another sound that had not been presented in the prime. This two-factorial manipulation resulted in four types of trials with respect to the prime-distractor–probe-target relation—that is, sound-repeated, location-repeated trials, sound-repeated, location-changed trials, sound-changed, location-repeated trials, and, finally, sound-changed, location-changed control trials in which nothing was repeated. Furthermore and on the basis of the findings in the Mayr et al. (2009b) study, we expected a pattern of results compatible with the feature mismatching hypothesis (Park & Kanwisher, 1994), as well as the object-binding framework—that is, slowed-down and/or more error-prone responding when the probe target sound mismatched the prime distractor sound but was presented at the same location (i.e., a sound-changed, location-repeated trial), relative to the sound-changed, location-changed control condition. On the basis of the object-binding framework, we also expected to find that, as compared with the sound-changed, location-changed control condition, responding on trials with repeated sounds but changed locations (i.e., a sound-repeated, location-changed trial) should be slowed down. In contrast, responding in the sound-repeated, location-repeated condition should not be impaired, as compared with the sound-changed, location-changed control condition, if mismatches were the sole determinant of auditory spatial negative priming; if the location and sound identity of the ignored prime distractor are repeated in the attended probe target, there can be no mismatch. If, however, distractor location inhibition or the retrieval of interfering prime episode information were also involved, responding in the sound-repeated, location-repeated condition should be impaired, relative to the sound-changed, location-changed control condition.

Method

Participants

The data of 5 participants had to be eliminated due to excessive error proportions ( > .50) in at least one condition. Four participants did not reach the learning criterion within the maximum number of training trials.The final sample of 75 participants (62 females) ranged in age from 19 to 35 years (M = 24). Participants were tested individually and either were paid or received course credit.

Materials

The stimuli were four digitized sounds (lamb, frog, bird, cat) of 300-ms length sampled at a rate of 44.1 kHz. The sounds differed considerably from one another with respect to spectral composition and were easily discriminable. Participants were seated in the center of a room, 60 cm in front of a 20-in. iMac. The computer was connected to four Yamaha MSP3 loudspeakers. Two speakers were placed in front of the participants, one 40° to the right of the sagittal head–computer-display axis, the other 40° to the left. Both speakers were placed at a 1.40-m distance from the participants. Two speakers were located in the same spatial arrangement, but behind the participants. Speakers were approximately at the height of the participants’ ears.

Each trial consisted of a prime and a probe display. Each display consisted of a target presented to one location and a simultaneously presented distractor at one of the other locations. A picture appearing on the monitor before each sound display indicated the to-be-attended sound. Participants reacted to the location of the attended sound by pressing one of the four keys on the number pad of the keyboard. The key layout was spatially compatible with the locations of the loudspeakers they represented (4, front left; 5, front right; 1, rear left; 2, rear right). The keys were pressed using the left and right middle and index fingers.

Prime–probe trials belonged to either the ignored repetition or the attended repetition design. In the ignored repetition design, the relation between the prime distractor and the probe target was systematically varied (as described below). In the attended repetition design, the relation between the prime target and the probe target was systematically varied. The ignored repetition design was central to the hypothesis tested, whereas the attended repetition design was primarily implemented to control for prime-to-probe response contingencies (in order to avoid predictability of probe responses based on prime responding).

For the ignored repetition design, half of the trials included a location repetition from the prime distractor to the probe target. For the other half of the trials, the location changed between the prime distractor and probe target. Sound identity was manipulated orthogonally to the location repetition manipulation. For 50% of the trials, the prime distractor sound was identical to the probe target sound; for 50% of the trials, the sound identity changed. As a consequence, 25% of the trials included a sound repetition and a location repetition between the prime distractor and probe target (sound-repeated, location-repeated), 25% included a sound repetition but no location repetition (sound-repeated, location-changed), and 25% included a location repetition only (sound-changed, location-repeated). For the remaining 25%, there was neither a location nor a sound repetition (sound-changed, location-changed; see Fig. 1 for an illustration of the four trial types).

In a parallel way, trials in the attended repetition design belonged to one of the four sound repetition (repeated vs. changed) × location repetition (repeated vs. changed) combinations, with the only difference that the repetition manipulation referred to the relation between the prime target and probe target. Note that 25% of the trials in each design (i.e., the sound-changed, location-changed trials)were entirely equivalent and did not include any prime-to-probe repetition; they can therefore be regarded as control trials against which the conditions with location and identity repetitions can be compared. Given the separate analyses of the two designs, we abstained from merging these two control conditions. Note, however, that additional analyses showed that response times and error rates did not differ between these two conditions when compared directly (the same was true in Experiment 2).

For a location-repeated trial in the ignored repetition design, prime and probe target and distractor locations were combined under two restrictions. The ignored prime location was identical to the attended probe location, and the attended prime location differed from the ignored probe location. For each location-repeated trial, a parallel location-changed trial was constructed by presenting the ignored prime sound at the only vacant location on the location-repeated trial. This first step resulted in a set of 24 trials, 12 location-repeated and 12 location-changed trials. This set of trials was quadruplicated.

Next, sound identity information was added. Each trial was instantiated once as a sound-repeated trial and once as a sound-changed trial. For each sound-repeated trial, the sound identities were randomly chosen, with the following restrictions. The prime distractor and probe target sound had to be identical, without any other sound repetitions. For each sound-changed trial, the sound identities were randomly chosen, with the restriction that they all had to differ from each other. In sum, the ignored repetition design comprised 192 trials, 48 in each location repetition (repeated vs. changed) × sound repetition (repeated vs. changed) condition. The construction was equivalent for the 192 trials of the attended repetition design. The 384 trials of the experiment were presented in a random sequence.

Procedure

Participants first adjusted the loudness to a comfortable level. On average, the loudness was set to 65 dB(A). Training trials were randomly chosen from the experimental trials. The experiment began only after 70% of the past 15 trials had been responded to correctly. If participants missed the criterion within 50 trials, they were offered the option of quitting or restarting the training.

Each experimental trial began with the visual picture cue of the sound that had to be localized subsequently. After a 500-ms interval, the two prime sounds were presented. Immediately after the prime reaction, the cue display was removed. A 1,000-ms prime–probe interval preceded onset of the visual cue for the probe. The two probe sounds were presented 500 ms later. Immediately after the probe reaction, the cue display was removed. In the prime as well as in the probe, reactions faster than 100 ms and slower than 3,000 ms were counted as invalid. Participants received audiovisual feedback about the correctness of their reactions. A 1,000-ms intertrial-interval preceded the next trial. After every 12th trial, participants received summary feedback. After the final trial, all participants were informed about the purpose of the experiment.

Design

The experiment comprised a 2 × 2 ignored repetition design with sound repetition (repeated vs. changed) and location repetition (repeated vs. changed) as within-subjects variables and another 2 × 2 attended repetition design with sound repetition (repeated vs. changed) and location repetition (repeated vs. changed) as within-subjects variables. The primary dependent variable was participants’ average reaction time, but error rates were also analyzed.

A power analysis was performed for the binary comparisons between the sound-changed, location-changed control trials, on the one side, and one of the other trial types, on the other. Assuming α = β = .05 and a population correlation of ρ = .3 in the reaction times between two to-be-compared conditions, effects of size f = 0.25 could be detected with N = 75 participants (Faul, Erdfelder, Lang, & Buchner, 2007). Whenever the statistical test of a substantive hypothesis implied a family of individual tests, the Bonferroni–Holm (Holm, 1979) method of protecting against α-error accumulation was applied in both of the experiments reported here.

Results

Probe reaction times were evaluated only for trials with correct prime and probe responses. For the ignored repetition design (Fig. 2), the location repetition manipulation was modulated by sound repetition with trials including identity–location mismatches—that is, sound-changed, location-repeated trials and sound-repeated, location-changed trials—being slowed down, as compared with control trials without a mismatch (i.e., sound-changed, location-changed trials). In contrast, sound-repeated, location-repeated trials were not slowed down, as compared with control trials.

The family of individual tests confirmed this pattern of results. As compared with sound-changed, location-changed control trials, sound-changed, location-repeated trials were significantly slowed down, t(74) = 3.76, p < .01, d _z = 0.43, as were sound-repeated, location-changed trials, t(74) = 2.80, p = .01, d _z = 0.32. In contrast, full repetitions (i.e., sound-repeated, location-repeated trials) did not differ from control trials (i.e., sound-changed, location-changed trials), t(74) = 0.02, p = .98, d _z < 0.01.

The analysis of the error data did not compromise the reaction time results, since there were no significant differences between any of the three comparisons, with t(74) = 1.91, p = .06, d _z = 0.22, for the comparison of sound-changed, location-repeated trials versus control trials, t(74) = −2.00, p = .05, d _z = 0.23, for the comparison of sound-repeated, location-changed trials versus control trials, and t(74) = −1.52, p = .13, d _z = 0.18, for the comparison of sound-repeated, location-repeated trials versus control trials.

Discussion

Experiment 1 was planned as a conceptual replication of the auditory location task reported by Mayr et al. (2009b), but with an extended stimulus set so that identity and location repetitions could be manipulated orthogonally. The present results nicely replicate and extend the findings of Mayr, Hauke, and Buchner.

First, responding was impaired, relative to the sound-changed, location-changed control condition, when there was a mismatch as originally conceived of by Park and Kanwisher (1994)—that is, when the probe target sound mismatched the prime distractor sound but was presented at the prime distractor location (i.e., a sound-changed, location-repeated trial). Second, responding was not impaired when there was no mismatch: Responses in the sound-repeated, location-repeated condition were as fast as responses in the sound-changed, location-changed control condition. This confirms our earlier conclusion that, in auditory space, neither inhibition of ignored locations nor retrieval of incompatible prime response information attached to the ignored location affects probe responding. This is in clear contrast to the visual modality, where impaired responding to previously ignored locations has been found irrespective of the stimulus identity at the repeated location (e.g., Milliken et al., 2000).

The present findings also extend those of Mayr et al. (2009b) by showing that responding not only was slowed down when the probe target sound mismatched the prime distractor sound but was presented at the prime distractor location. Rather, responding also was slowed down when the probe target was identical to the prime distractor but appeared at a different location. In sum, then, we may conclude that auditory spatial negative priming occurs whenever there is a mismatch between prime and probe displays but does not occur when there is no mismatch, even though the probe target is a complete sound and location repetition of the prime distractor and, thus, should invite distractor inhibition, the retrieval of incompatible prime response information, or both.

Slowed-down responding in Experiment 1 entirely depended on the presence of a feature mismatch of any kind. This finding fits well within the object-binding framework. Auditory features (in a similar way as visual features) are spontaneously integrated into episodic auditory object files, which are stored in memory (Zmigrod & Hommel, 2009, 2010). These episodes are assumed to be retrieved as a whole when at least one of the elements is encountered again on a later trial. Retrieval does not take place when there is no repetition of any of the features. This would be equivalent to our sound-changed, location-changed control trials. Object file retrieval is unproblematic or even beneficial for subsequent responding if all features bound together are repeated—that is, in a full repetition trial. This was the case in our sound-repeated, location-repeated trials. However, partial repetitions (i.e., repetitions of some but not all features) lead to reactivation of misleading and conflicting information, which slows down processing and responding. This was true for both of our mismatch trials, the sound-changed, location-repeated trials and the sound-repeated, location-changed trials. In sum, then, our findings argue for the validity of the object-binding framework, but with the extension that object binding occurs not only for task-relevant and task-irrelevant features of a single task-relevant object, but also for features of a totally irrelevant and interfering object that should have been ignored. Thus, the mechanisms underlying stimulus feature integration seem to be quite potent.

Similarly, at a conceptual level, but with an auditory recognition memory task, Maybery et al. (2009) presented evidence of auditory feature binding across identity and spatial information. They used sequences of auditory events that varied with respect to identity (e.g., spoken letters) and location (played from one of eight loudspeakers). A single test sound was presented from a single loudspeaker. Participants had to classify the test sound as old when it was a sound from the study sequence delivered from its original location (intact probe) or when it was a sound from the study sequence but delivered from a location that had been occupied by another study sound (recombined probe). “New” test probes comprised either a sound or a location (or both) not presented in the study sequence. If auditory identity and location information were stored separately in working memory, recognition performance for intact and recombined probes should be identical. However, it turned out that intact probes were recognized better than recombined probes, suggesting that auditory objects are stored as integrated sound identity–location representations.

The findings reported by Maybery et al. (2009) and by Zmigrod and Hommel (2009) and those of the present Experiment 1 are all demonstrations of auditory object integration of identity and spatial information. However, auditory object integration seems to be complicated by at least two aspects. First, Zmigrod and Hommel (2009) showed that features associated with task-relevant stimulus dimensions are more likely to become part of bindings than are task-irrelevant features, suggesting that attention is beneficial to the integration process. Second, there seems to be an asymmetry between the different feature dimensions in audition, in that identity information may be predominant and may carry more weight than spatial information. For instance, Maybery et al. reported that it was easier to correctly classify as new an auditory event composed of a new sound identity and an old location than an auditory event composed of a studied sound identity at a new location. Obviously, the change of a sound’s identity was more important in judging an item as new than was a change of its location. This fits well with Kubovy and Van Valkenburg’s (2001) conclusion that “space is not central to the formation of auditory objects” (p. 99) but that auditory object formation is based on nonspatial properties. If the nonspatial property (such as the frequency distribution) changes, the object is no longer the same, irrespective of location repetition or change, but when the nonspatial properties remain, location changes do not impair the impression of perceiving the same object.

The asymmetry between identity and spatial information processing in auditory working memory became even more apparent when Maybery et al. (2009) slightly changed the task in their Experiments 3 and 4. Participants had to decide whether the recognition probe was an old item with respect to sound identity (verbal recognition task) or with respect to sound location (spatial recognition task). The results were task specific. Participants in the spatial recognition condition were again better at recognizing intact than recombined probes. In contrast, participants in the identity recognition condition recognized intact and recombined probes equally well. The authors concluded from this pattern of results that identity is a more fundamental dimension than spatial location in the representation of auditory objects and, as a consequence, that sound sequences could be encoded and stored with reference to their identity without including a reference to their location (Maybery et al., 2009). In contrast, it may not be possible to encode and store spatial sound information without reference to the identity of the sound. In other words, encoding sound identity information seems to be obligatory, whereas encoding location information is not. Whenever location information is central to a task, such as in the spatial recognition task, identity information is mandatorily encoded and retrieved as well, so that evidence of feature binding can be found. In contrast, when identity information is central to the task, as in the verbal recognition task for which location is irrelevant, location information is not necessarily encoded and/or retrieved, so there will be no evidence of feature binding.

Returning to the starting point of these considerations—that is, to the auditory location negative priming task used in the present Experiment 1—location information processing was obligatory because determining the location of a sound was essential to performing the task. Following Maybery et al. (2009), obligatory auditory location processing implies encoding sound identity information as well. It suggests itself that this is why the partial repetition costs in Experiment 1 were symmetrical: Both types of mismatches—changes of the location of a repeated sound and changes of the sound identity at a repeated location—resulted in reduced performance. If these considerations are valid, the partial repetition costs may be asymmetrical when the task of locating a sound of a certain identity is replaced by the task of determining the identity of a sound at a certain location, as in an auditory identity negative priming task. Here, sound identity is central to performing the task, but location is not. Therefore, the repetition of a sound identity, but at another location, should be perceived as sound repetition with mismatching location information. In contrast, the repetition of a location occupied by another sound identity may stay undetected. In other words, for an auditory identity negative priming task, we would expect a performance decrement, relative to the sound-changed, location-changed control condition, in the sound-repeated, location-changed condition, but not in the sound-changed, location-repeated condition.

Identity negative priming and feature mismatching

While this expectation regarding performance in an identity priming task may seem reasonable within the present context, it also appears to contradict the available empirical evidence, which seems to show quite clearly that identity–location mismatch does not play a role in auditory identity negative priming, (1) in that the size of the negative priming effect is equivalent for trials with and without a mismatch between prime and probe and (2) in that auditory negative priming can be observed in the absence of a mismatch (Buchner & Mayr, 2004; Buchner & Steffens, 2001; Mayr & Buchner, 2006, 2010a, 2010b; Mayr et al., 2009a, 2010b; Mayr, Niedeggen, Buchner, & Orgs, 2006; Mondor et al., 2005).^{Footnote 2} How can this apparent contradiction be resolved? In short, we assume that identity–location mismatches may not have been observed in earlier auditory negative priming experiments because the sounds in those experiments had no natural location in three-dimensional space. Instead, sounds were dichotically presented over headphones, as is common practice in selective attention experiments, such that “location” was actually equivalent to the ear of entry. In such a setup, natural cues from which a sound’s position in space can be computed by the auditory system either were not available at all (interaural time differences, head-related transfer functions, distance cues) or were available in a format that never occurs in real-world situations (interaural level differences: full level at one ear, zero at the other). Without any natural cues to position in space, the location feature in an auditory object file could not be computed and may thus have been missing, as a consequence of which an identity–location mismatch could not be observed.

In order to test this possibility, we decided to use an auditory identity priming task with the same four-location auditory free-field setup as in Experiment 1. To maximize the comparability between experiments, we also used the identical auditory stimuli, the same trial setup comprising an ignored repetition and an attended repetition design, the same response mode, and endogenous visual cues that indicated the to-be-attended location for the next display.^{Footnote 3}On the basis of the findings in Experiment 1, the object-binding framework, and the particular significance of space in audition, we expected to find a slowdown for sound-repeated, location-changed mismatch trials, as compared with sound-changed, location-changed control trials. In contrast, on the basis of our previous findings in auditory identity negative priming, we did not expect feature mismatching to be of importance. Instead, the sound-repeated, location-changed trials, as well as the sound-repeated, location-repeated trials, should be slowed down, as compared with sound-changed, location-changed control trials, because the former included the repetition of a previously ignored sound identity.