Stimulus-dependent processing of temporal order

Abstract

Two distinct conceptualisations of processing mechanisms have been proposed in the research on the perception of temporal order, one that assumes a central-timing mechanism that is involved in the detection of temporal order independent of modality and stimulus type, another one assuming feature-specific mechanisms that are dependent on stimulus properties. In the present study, four different temporal-order judgement tasks were compared to test these two conceptualisations, that is, to determine whether common processes underlie temporal-order thresholds over different modalities and stimulus types or whether distinct processes are related to each task. Measurements varied regarding modality (visual and auditory) and stimulus properties (auditory modality: clicks and tones; visual modality: colour and position). Results indicate that the click and the tone paradigm, as well as the colour and position paradigm, correlate with each other. Besides these intra-modal relationships, cross-modal correlations show dependencies between the click, the colour and the position tasks. Both processing mechanisms seem to influence the detection of temporal order. While two different tones are integrated and processed by a more independent, possibly feature-specific mechanism, a more central, modality-independent timing mechanism contributes to the click, colour and position condition.

Introduction

Two main concepts constitute our experience of time: succession and duration (Fraisse, 1984, Wittmann, 1999). Perception of succession refers to the sequential characteristics of events, i.e. their temporal order. Perception of duration refers to the time interval between two events or to the persistence of an event over time. Numerous studies on humans assessing the ability to perceive the temporal order of successive events have been conducted over the last decades. As temporal processing of order is discussed as a basis for higher cognitive abilities, such as speech and language processing (Tallal et al., 1998, Wittmann and Fink, 2004, Wittmann et al., 2004), the identification of these processing mechanisms becomes an important issue in clinical neuropsychology.

Several studies have shown that the perception of the temporal order of acoustic events is only possible for humans when the events are separated by inter-stimulus intervals (ISI) of about 20–40 ms (e.g. Hirsh, 1959, Lotze et al., 1999, Pastore and Farrington, 1996, Kanabus et al., 2002). Moreover, temporal-order thresholds have been assessed over different sensory modalities, and results have shown approximately the same values for auditory, visual and tactile stimulation (Hirsh and Sherrick, 1961). In the auditory modality, the stimuli varied either in pitch, in location (right ear or left ear) or in a combination of pitch and location. In the visual modality, two flashes of lights were presented at different locations. Results showed no significant differences in temporal-order thresholds related to different stimulus parameters or sensory modalities.

These results point to a central-timing mechanism responsible for temporal-order judgement (Pöppel, 1997). This mechanism is thought to create basic temporal units of perception and provide discontinuous processing of information based on neuronal oscillations in the 40-Hz range (Joliot et al., 1994, Pöppel, 1970). Every time a stimulus is processed in a specific modality, a neuronal oscillation is initiated. Each oscillation has a period of about 30 ms and represents one processing unit. Within these processing units, incoming events are treated as co-temporal (Pöppel et al., 1990). The temporal order of sensory events cannot be indicated when the two events occur during one oscillation period. Therefore, the temporal order of stimuli which are separated by less than 20–30 ms cannot be detected. It is assumed that these system states allow the integration of information from different sensory modalities and provide a logistical basis for the identification of the temporal order of events.

Various experiments provide neurophysiological correlates for such discrete temporal-processing stages. Magnetoencephalographic recordings (MEG) in humans suggest that 40-Hz activity is involved in the perceptual separation of acoustic events over time. These MEG recordings are believed to result from a coherent 40-Hz resonance between thalamocortical loops responsible for binding neural processes involved in perception (Joliot et al., 1994, Llinas and Ribary, 1993). Rhythmic brain activity in the 40-Hz range (the gamma band) is thought to provide fundamental temporal-building blocks in sensory and cognitive processing (Gray et al., 1989, Schwender et al., 1994, Basar-Eroglu et al., 1996).

Some psychophysical studies have shown the influence of the applied experimental procedure on threshold size (Lotze et al., 1999, Mills and Rollman, 1980, Pastore et al., 1982). Components that are considered to contribute to differences in temporal-order thresholds are spatial location of sensory stimuli (Swisher and Hirsh, 1972) and directed attention (Gibson and Egeth, 1994, Stelmach and Herdman, 1991). For example, in experiments investigating the effect of directed attention, subjects’ attention is allocated to one of two stimuli. Results show that stimuli on which attention is focused are perceived to occur before the other stimulus when the two visual stimuli are presented at the same time (Stelmach and Herdman, 1991).

This effect, however, disappears when a three-response paradigm is employed (Jaskowski, 1993). In an experimental set-up, two visual stimuli were presented positioned one above the other, and two different paradigms were compared: a two-response paradigm (possible answers: top first, bottom first) and a three-response paradigm (possible answers: top first, bottom first, simultaneous). The results of the two-response paradigm confirmed the effect of directed attention. In the three-response condition, however, no effect of allocating the attention to one stimulus was found. Jaskowski concludes that in two-response paradigms responses are biased when the subjects are forced to make a decision, although they cannot identify the temporal order of the stimuli. In these situations, they decide in favour of the stimulus to which they directed their attention. This bias disappears when the subjects can give the answer “simultaneous”.

As McFarland et al. (1998) demonstrated, temporal-order thresholds are significantly influenced by the physical properties of the stimuli employed. In their study, auditory stimuli varied either in frequency or sound-pressure level, and visual stimuli differed in size, orientation or colour. Results showed that temporal-order thresholds differ significantly over the various stimulus dimensions within the auditory and visual modalities. Thresholds were lowest for pitch differences in the auditory domain. Higher temporal-order thresholds were found for the stimulus dimensions sound-pressure level in the auditory domain and size and orientation in the visual domain. Highest thresholds occurred for the colour paradigm in the visual domain. The authors assume that these differences are due to specific perceptual mechanisms within the sensory domains. Lower thresholds for size and orientation, as compared with colour, can be explained in terms of apparent motion cues. Changes from large to small rectangles and vice versa in the size condition can result in the impression of motion (impression that objects move towards or away from the subject). Variations of two orthogonal positions of bars (vertical and horizontal in the orientation condition) can result in the perception of radial motion. These percepts can serve as additional cues to discriminate the order of the stimuli at short intervals. Thus, these results contradict the hypothesis of a central-timing mechanism and suggest feature-specific discrimination mechanisms.

To sum up, two different conceptualisations of temporal-order processing have been proposed. One assumes a central-timing mechanism involved in temporal-order judgement. The other suggests feature-specific discrimination mechanisms. As discussed above, in certain stimuli the integration of subsequently-presented features to one percept can serve as an additional cue for the determination of temporal order. In the visual modality, the rapid serial appearance of two stimuli at different positions gives the impression of apparent motion (Exner, 1875), which activates neurons in motion-sensitive areas of the brain (Newsome et al., 1986, Seiffert et al., 2003). The temporal order of certain stimuli can thus be reconstructed from the motion percept without involvement of a central timing mechanism and, therefore, also without its temporal limitations. The mechanisms of motion detectors have actually been described without recourse to a central timing mechanism (Adelson and Bergen, 1985, Reichardt, 1957). Similar stimulus-specific processing mechanisms are also found for auditory stimuli (changes in pitch). In the primary auditory cortex, neurons are selective for the direction of frequency-modulated sounds. The two tones are integrated into one percept with different spectral patterns (high-to-low; low-to-high) according to the order presented (McKenna et al., 1989, Rauschecker, 1998, Tian and Rauschecker, 1998).

The present study was performed to investigate the relationships between four temporal-order tasks conducted in the auditory and visual domain. These tasks are frequently employed in clinical neuropsychology and often treated as related to common temporal-processing mechanisms. Thus, the aim was to draw conclusions about whether there are common processes underlying temporal-order thresholds over different modalities and stimuli or whether distinct processes are triggered by each task. The following stimulus settings were used for assessing perception of temporal order: in the auditory domain, a pair of clicks, one to each ear, was presented (task 1) and a pair of tones with different frequencies was presented to both ears (task 2). In the visual domain, two stimuli with different colour appeared at the same position (task 3) and two stimuli with identical physical properties appeared at different positions (task 4). Based on earlier studies described above, we assumed that the presentation of two tones with different frequencies in the auditory domain and the presentation of two stimuli at different locations in the visual domain are likely to trigger cue-specific mechanisms for the detection of temporal order as opposed to the click stimuli and the colour stimuli, respectively. Therefore, different absolute order-threshold values are expected. In addition, to evaluate whether there are common processes underlying the four measures, correlations and regression analyses were performed. However, instead of indicating common underlying temporal mechanisms, positive correlations between different temporal-order tasks can be explained by common mediating influences such as the attentional performance of the subjects or their working-memory capacities. To evaluate this possibility, additionally, alertness, divided attention, as well as verbal and spatial working memory span were assessed and controlled for using partial correlations.