An essential prerequisite for human behavior is the ability to actively maintain representations of information that is no longer available, flexibly update and reorganize this information and use it for guiding actions or thoughts. All these characteristics are subsumed under the psychological construct of working memory (WM) which is assumed to have a limited capacity and operate in a time frame up to several seconds (Baddeley, 03; Fuster, 00; Levy und Goldman-Rakic, 00). WM has been mainly investigated in the visual and auditory modality. However, WM for tactile features is also important for everyday behaviors such as manual object recognition: Imagine you have to find your apartment key in a fully packed handbag without looking. First, it is necessary to activate a mental representation of the tactile features of your key. Then you have to search with your hand through different objects including other very similar keys while maintaining the representation. Finally, you have to recognize the particular key you need. Because tasks like this require complex representations of different stimulus features even within the somatosensory modality and involve a multitude of different processes, a first step in the investigation of tactile WM is to focus on a sub-modality of the somatosensory system and to use a relatively simple sensory WM task. The aim of the present dissertation was to investigate the neural organization of WM for vibratory flutter stimuli in the healthy human brain. For this purpose, a vibrotactile delayed discrimination task was studied using methods of functional magnetic resonance imaging (fMRI), electroencephalography (EEG), concurrent subliminal electrical stimulation and psychophysics.
There are many psychological theories about WM and a complete review of these different theories is beyond the scope of the present dissertation (see Miyake and Shah for an overview (99)). Here, only two models are briefly summarized that indicate the range of psychological WM theories and are pertinent for the present dissertation. One influential model that emphasizes the dissociation of maintenance of information and executive processes in WM was developed by Baddeley and Hitch (Baddeley und Hitch, 74; Baddeley, 03; Repovs und Baddeley, 06). This model proposes the existence of multiple specialized compartments which interactively enable WM. A central executive is assumed to control the operation of modality-specific storage systems, the phonological loop and the visuo-spatial sketchpad, which actively maintain memory representations for a limited period of time. Although this model only deals with the storage and processing of phonological and visuo-spatial information, the authors assume that there must also be modality-specific storage systems for the other modalities such as tactile information. Another model was proposed by Cowan (93; 88) who regards WM as the subset of activated representations from long-term memory (LTM). He suggests that a subset of these activated memory representations are in the focus of attention and conscious awareness. The direction of attention is controlled voluntarily by a central executive or automatically by an attentional orientation system initiated by salient or significant stimuli. Because only 3 to 5 unrelated items can be in the focus of attention simultaneously and the activation of representations fades over time, WM capacity is limited.
Besides all theoretical differences, these WM models illustrate important aspects of WM. First, WM involves the active maintenance of information. This can either be information perceived from the external world or re-activated internal representations. Importantly, the temporarily maintained information is actively protected against decay and from interfering or distracting neural activity. Second, WM is subject to capacity limits regarding the amount of information and the time for which information can be maintained. Third, WM does not only involve the maintenance of these representations but engages executive processes and attention, which serve to re-organize or manipulate the maintained information and determine which information enters WM. Fourth, WM provides a necessary basis for goal-directed behavior and complex cognition. In addition, whereas the model of Baddeley and Hitch stresses the importance of modality-specific storage systems, Cowan's model emphasizes the importance of LTM representations for WM.
For phonological and visuo-spatial information, the dissociation between storage and executive components of WM was confirmed in neuroimaging studies and in studies with brain-damaged patients who suffered from specific neuropsychological symptoms (for review see (Baddeley, 03; Curtis und D'Esposito, 03; Smith und Jonides, 98)). In general, these studies suggest that the storage component of WM is located in those posterior modality-specific sensory or association areas that are also involved in the on-line processing of a specific type of sensory information. Executive processes on the other hand appear to be associated with the activity of brain areas distinct from sensory areas, most importantly the lateral prefrontal cortex (PFC).
Single-unit recording studies to explore the neural correlates of WM in non-human primates used delayed sensory or response tasks, which focus on WM maintenance and place relatively little demand on executive processes. In these studies, stimulus-selective persistent activity during the delay period of WM tasks is considered to be the neurophysiological correlate of stimulus representations maintained in WM (Hebb, 49; Goldman-Rakic, 95). Similar to the findings in human subjects, stimulus-selective delay period activity has been observed in modality-specific brain regions of the posterior association cortex (Chafee und Goldman-Rakic, 98; Fuster und Jervey, 81b). In monkeys, however, stimulus-selective activity has also been found in the lateral PFC in delayed-response (Funahashi et al., 93) and delayed matching or discrimination tasks (Sawaguchi und Yamane, 99; Miller et al., 96b) suggesting that this region is actively involved in WM maintenance. Importantly, delay-activity was reduced and performance was impaired by cooling-induced transient lesions of the lateral PFC (Chafee und Goldman-Rakic, 00; Fuster et al., 85).
Both, human neuroimaging and non-human single-unit studies, have shown that the PFC seems to be the only brain area with the ability to actively maintain memory representations that survive the appearance of intervening or distracting stimuli (Miller et al., 96; Sakai et al., 02a). In addition, the studies in non-human primates suggested a material-specific organization of the lateral PFC with its dorsolateral part associated with the maintenance of visuo-spatial material and its ventrolateral part linked to the maintenance of non-spatial object information (Levy und Goldman-Rakic, 00). These findings agree with anatomical findings showing that modality-specific posterior brain regions are interconnected with distinct sub-regions of the PFC (Petrides und Pandya, 99; Petrides und Pandya, 02; Barbas, 92). They also suggest that modality-specific networks including posterior association areas and lateral PFC conjointly enable the active maintenance of information in WM. Contrary to these findings, results of neuroimaging studies found no evidence of a sub-specialization of the lateral PFC according to the material processed but instead lead to the proposal of a process-specific organization (Frith et al., 91; D'Esposito et al., 00; Petrides, 05).
Although the lateral PFC seems to be crucial for WM, its exact role in the maintenance of information is still under debate (Curtis und D'Esposito, 03b; Postle, 05). Humans with lesions of the lateral PFC are relatively unimpaired in delayed sensory or motor tasks (D'Esposito und Postle, 99). Furthermore, it remains unclear what the observed stimulus-selective persistent activity in the lateral PFC reflects. A recent study demonstrated that the majority of task-specific neurons in the lateral PFC are tuned to selective attention to a specific location as opposed to memory for the location (Lebedev et al., 04). Some authors therefore disagree with the view that stimulus representations are maintained in the lateral PFC and assume that persistent activity in this area reflects processes assisting in maintenance such as top-down control (Miller, 00), selective attention (Lebedev et al., 04) or response selection (Rowe und Passingham, 01).
Rather than supporting pure maintenance, it has been suggested that the lateral PFC modulates processing in posterior brain regions and selectively activates task-relevant representations (Duncan und Owen, 00; Kimberg und Farah, 93; Miller, 00; Curtis und D'Esposito, 03; Fuster, 00). Maintenance of stimulus representations is instead supported by a parieto-premotor cortical network (Manoach et al., 03; Rowe et al., 00; Sakai et al., 02) and modality-specific sensory association areas (Pasternak und Greenlee, 05). These authors, therefore, emphasize the function of the lateral PFC in executive control. However, the existence of stimulus-specific delay activity in lateral PFC, which has been observed in various experiments using different stimulus modalities, suggests that the lateral PFC is also involved in WM maintenance, at least in non-human primates. Species differences in brain function might be one reason for the inconsistent findings regarding the role of the lateral PFC in WM. The PFC is the brain region that underwent the proportionally largest increase in the evolution from non-human primates to humans implicating a significant role in complex human cognition (Fuster, 02).
On the other hand, psychological accounts, human neuroimaging studies and single-unit recordings in non-human primates approach the phenomenon of WM from different perspectives. Psychological theories try to provide categorical descriptions of cognitive processes related to observable behavior. Previous approaches for understanding the neural basis of WM might be biased by the attempt to localize theoretically defined psychological constructs in specific brain regions. It is especially tempting to do this when using the method of fMRI. However, processes are implemented in the brain by neural networks and not located in single regions. Furthermore, neuroimaging studies rely on the method of cognitive subtraction and might therefore be more sensitive to specific cognitive processes (Friston et al., 96). In addition, single-unit and neuroimaging studies have a different spatial resolution: whereas neuroimaging studies provide a granular indicator of neuronal activity, single-unit studies can detect stimulus-selective activity in single neurons or populations of neurons. As pointed out by Repovs and Baddeley (06), the different levels of description cannot (yet) be mapped in a 1-to-1 manner but they provide insight into the phenomenon of WM from different perspectives. Maybe new approaches that regard WM as an emergent property of the functional interaction between networks of brain areas provide a promising perspective (D'Esposito, 07; Postle, 06).
The functional interaction between brain regions can only be investigated by exploiting the advantages of different methods including complementary studies in humans and monkeys. In addition, to get a more complete picture of cognitive functions and their neural basis, they have to be studied from different perspectives including psychological and neuroscientific approaches. Therefore, in the present dissertation the same vibrotactile WM that was used in previous studies in non-human primates was investigated with a multi-method approach using fMRI, EEG and behavioral studies in humans.
The above mentioned studies investigated WM for visuo-spatial, visuo-object, phonological or auditory information. However, only a few studies addressed WM for tactile features which will be further investigated in the present dissertation. According to the emergent property view of cognitive processes, WM for a specific type of sensory information must be implemented in the brain according to the neuroanatomy, physiology, and connections of the modality-specific regions it operates on. As the underlying research of this dissertation refers to somatosensory WM, the neuroanatomy of the somatosensory system will be briefly introduced in the following section before findings regarding vibrotactile WM are discussed.
The somatosensory system conveys information about the outside world that is transmitted via the skin and information about the body itself (for review see (Kaas, 93; Kaas, 04)). It comprises four distinct modalities. Proprioception is caused by mechanical displacement of receptors in skin, muscles and joints. Pain is produced by noxious stimuli in free nerve endings. Thermal sensation is generated by cold and warm stimuli. Touch is elicited by the mechanoreceptors of the skin which can be divided into two functional groups. This dissertation only deals with the modality of touch. In touch, rapidly adapting mechanoreceptors, i.e., Meissner and Pacini corpuscles, respond to the onset and offset of a stimulus. They are located in the hairless skin and because of their small receptive fields (2-4 mm) can resolve fine spatial differences. Slowly adapting receptors, i.e., Merkel and Ruffini corpuscles, respond to persistent stimuli. They are located in the subcutaneous tissue and have much larger receptive fields (5 cm). Depending on the specific stimulus features, a tactile sensation is usually a result of a combined activation of different receptor types. In the present dissertation vibrotactile stimuli in the range of 10 to 43 Hz were used. These stimuli have been shown to primarily activate Meissner receptors and elicit a sensation called flutter (Mountcastle et al., 67).
The sensory information from the mechanoreceptors is transmitted to the brain via the dorsal column–medial lemniscal pathway (see Figure 1a): over the dorsal column nuclei the information reaches the spinal cord, and is then transmitted via the dorsal column pathway to the medulla oblongata where the fibers cross to the other side. The information is then transferred via the medial lemniscial pathway to the ventroposterior lateral (VPL) and the ventroposterior medial (VPM) nuclei of the thalamus. From the thalamus, the signal is directly projected to the primary somatosensory cortex (S1) and to a lesser extent directly to the secondary somatosensory cortex (S2).
|Figure 1. The somatosensory system. a) The somatosensory pathway from the mechanoreceptors to the primary (S1) and secondary (S2) somatosensory cortex. b) S1 comprises the postcentral gyrus in the anterior parietal lobe and contains the Brodman areas (BA) 3b, 3a, 1, 2 (see enlarged cutout). S1 projects to the ipsi- and contralateral S2, the primary motor cortex (M1), the inferior (IPL) and superior posterior parietal lobe (SPL). S2 is located in the upper bank of the Sylvian fissure and projects to contralateral S2, IPL, SPL, the ventral (vPMC) and medial premotor cortex (mPMC) and lateral prefrontal cortex (PFC).|
S1 is located in the postcentral gyrus of the anterior parietal lobe and consists of four cytoarchitectonically distinct regions (see Figure 1b): Brodmann areas (BA) 3a, 3b, 1 and 2. The subregions of S1 are functionally specialized: whereas neurons in 3a and 2 respond predominantly to proprioceptive stimulation, neurons in BA 3b and 1 are primarily activated by mechanical stimulation. In primates, BA 3b contains the largest representation of the finger with the finest resolution (Blake et al., 02). S1 neurons primarily represent physical properties of stimuli.
Most thalamic projections end in BA 3b and 3a. BA 1 and 2 receive much less direct thalamic inputs but receive their main input from BA 3a and 3b. Each of the S1 subregions contains a complete somatotopic map of the entire contralateral body. Recently, evidence for an ipsilateral representation has been found (Iwamura, 00; Palva et al., 05). Body parts that are more sensitive to touch (e.g., fingers and lips) are represented by a relatively larger area of this map resulting in the so-called homunculus. S1 is reciprocally connected with S2 and the primary motor cortex (M1) (Kaas, 93).
S2 is located in the upper bank of the Sylvian fissure (see Figure 1b) and smaller in size than S1. S2 neurons have receptive fields which are larger and bilateral. S2 exhibits a somatotopical respresentation of the ipsi- and contralateral body but with much less detail than S1. S2 is reciprocally connected with the ipsilateral S1 and via the corpus callosum with the contralateral S1and S2. S2 receives a small direct input from the VPL. S2 projects to inferior and superior posterior parietal cortex (PPC) and lateral PFC (Cipolloni und Pandya, 99; Carmichael und Price, 95) (see Figure 1b). PPC contains neurons with tactile and multimodal receptive fields and projects to the premotor cortex (PMC), the lateral PFC and the insular cortex (Friedman et al., 86; Kaas, 93; Petrides und Pandya, 84; Breveglieri et al., 06; Geyer, 04; Seltzer und Pandya, 80; Preuss und Goldman-Rakic, 89). Most experimental data hint towards a serial processing of information from S1 to S2 (Allison et al., 89a; Allison et al., 89b; Hari et al., 84). However, some studies were able to demonstrate the existence of parallel direct pathways to S1 and S2 (Barba et al., 02; Karhu und Tesche, 99). Whereas lesions of S1 impair the localization and detection of tactile stimuli, lesions of S2 lead to more complex impairments including disturbed shape and roughness processing, temperature perception, and nociception (Freund, 03; Duncan und Albanese, 03). PPC is implicated in even more complex tactile processing including attention and memory, multi-modal processing and somato-motor programming (Burton und Sinclair, 00; Jeannerod et al., 94; Driver und Vuilleumier, 01). Lesions of PPC can cause the syndrome of tactile neglect and apraxia (Vallar et al., 03; Binkofski et al., 01).
Several studies in monkeys and humans investigated WM for vibrotactile information in the range of flutter frequencies. Romo and colleagues (Hernandez et al., 00; Hernandez et al., 02; Romo et al., 99; Romo et al., 02b; Romo et al., 02a; Romo und Salinas, 03; Romo et al., 04; Salinas et al., 00) conducted a series of studies in which neural activity from single units in S1, S2, PFC, medial and ventral PMC was recorded while monkeys were performing a vibrotactile delayed discrimination task. A similar version of their task was used in all experiments of the present dissertation. In the vibrotactile delayed discrimination task, two vibrotactile stimuli differing in vibration frequency were applied to the monkeys' index finger separated by a delay ranging from 6 to 8 s. After the second vibration was applied, the monkeys had to indicate which of the two vibrations had the higher frequency. This task comprises different epochs: encoding of the vibrotactile stimulus, maintenance of this stimulus over a delay period and finally decision making comprising the comparison of the first with the second vibrotactile stimulus.
The authors found that the majority of S1 neurons encoded the frequency of the vibrotactile stimuli in the periodicity of the firing pattern but about one third of S1 neurons modulated their firing rate as a function of the stimulus frequency. The firing rate of these later neurons was parametrically modulated by the frequency of the vibrotactile stimulus, i.e., the neurons fired more frequently with increasing vibration frequency. Discrimination performance only decreased slightly when aperiodic vibrations with irregular intervals between single pulses were applied. This finding indicates that the periodicity of spikes cannot be the necessary code for performing the task. Even when electrical current pulses with different frequencies were used to intracranially stimulate S1 instead of applying peripheral vibrotactile stimulation, monkeys still correctly performed the delayed discrimination (Romo et al., 00). Moreover, S1 neurons in the macaque brain show higher firing rates during active discrimination than during passive vibrotactile stimulation (Salinas et al., 00). Together, these results show that the neural representation necessary and sufficient to perform the vibrotactile delayed discrimination task is a firing rate code that parametrically encodes the frequency of the vibration stimulus. In addition to S1, neural activity in S2, lateral PFC, medial and ventral PFC also parametrically represents the vibration frequency during the presentation of the first and the second vibration.
During the maintenance period of the task, Romo and colleagues found neurons that varied their firing rate parametrically depending on the frequency of the first vibration in S2, lateral PFC, ventral and medial PMC. Notably, no delay activity was found in S1. Whereas in S2 only neurons with early delay activity were found, delay-related neurons in PMC only showed stimulus-modulated firing rates in the late delay period. Lateral PFC was the only region where neurons with sustained firing rates during the entire delay period were observed. Interestingly, while S1 neurons only exhibited positively monotonic tuning curves, the firing rate of the delay-related neurons in these other regions increased or decreased as a monotonic function of the frequency of the first vibration. For S2 neurons it has been shown that the existence of two populations of neurons with opposite tuning functions enhances the fidelity of the neural representation by reducing noise and enhancing signal strength (Romo et al., 03). Periodicity was almost absent in S2 and could not be identified in PFC or PMC. Similarly to S1, sustained firing in these regions was reduced during passive stimulation and the firing rate elicited by the second vibration stimulus correlated with behavioral performance. Taken together, these findings suggest that S2, PFC and PMC are part of the neural circuitry mediating the maintenance of the vibrotactile memory trace whereas S1 seems to generate the neural representation of the stimulus, i.e., translating a periodic into a firing rate code. Furthermore, the physiological basis of the somatosensory memory trace seems to be represented parametrically by the firing rate of neurons and not categorically. It is important to note that this was the first time that a firing rate code had been linked to a sensory memory representation.
Neurons in S2, lateral PFC, medial and ventral PFC exhibit activity patterns that reflect the evolution of the behavioral decision. During the presentation of the second vibration some neurons responded at a rate that exclusively reflected the frequency of the first or the second vibration. However, the majority of neurons modulated their response according to the difference between the first and the second vibration. Accordingly, about half of these neurons fired with a higher rate when the first stimulus had a higher frequency than the second whereas the other half showed the opposite pattern. The sign of this difference correlates with the monkeys' behavioral response indicating that the observed modulations of firing rate are indeed functionally relevant. Together, these findings indicate that the firing pattern of neurons in these brain regions becomes gradually correlated with the monkey choice during the presentation of the second vibration. This indicates that these neurons reflect and are crucially involved in the comparison process gradually evolving into the decision. The outcome of this decision-making process is transferred to M1 which generates the overt behavioral response. The decision related activity occurs earliest in ventral PMC followed by PFC, medial PMC and then in S2. Therefore, it has been suggested that the decision process in this task might be initiated and controlled by the ventral PMC, continued and amplified by lateral PFC and medial PMC, and its result sent as an "efference copy" to S2. The finding that two oppositely tuned populations of neurons integrate neural evidence over time leading to a final decision is similar to results of single-unit recordings in PPC and PFC of monkeys performing a visual discrimination task (Kim und Shadlen, 99; Shadlen und Newsome, 01).
In summary, theses studies imply that a cortical network of somatosensory, prefrontal and motor areas provides the neural basis necessary for performing the vibrotactile delayed discrimination task. These studies revealed three important aspects related to the neural basis of cognitive processes. First, performance is conjointly realized by a distributed neural network. Except for S1 and M1, activity of neurons reflects the entire sequence of processing steps, i.e., encoding, maintenance and decision making that link sensation and action in the task. Second, different conceptually defined cognitive processes (e.g., encoding, maintenance, decision making) are not located in distinct brain regions. Specialization of a specific brain region is in fact a result of the relative strength with which it contributes to specific process. Third, neural activity does not proceed in separate steps but develops gradually in the involved brain regions to enable task performance.
Although these single-unit studies revealed important insight regarding the neural implementation of vibrotactile WM, brain activity was only studied in a few pre-selected brain regions. Therefore, it remains an open question which additional brain regions are part of the neural network supporting vibrotactile WM. For instance, in the visual modality it has been shown that the sustained maintenance of information and decision making is implemented in posterior modality-specific association areas (Pasternak und Greenlee, 05; Gold und Shadlen, 07). However, Romo and colleagues did not record in the PPC which represents an important association area of the somatosensory system. In the present dissertation, fMRI was used to study the whole brain when subjects performed a vibrotactile delayed discrimination task (Study I).
Comparative psychophysical studies in humans and monkeys have shown that both species have similar abilities to discriminate between vibrotactile frequencies (Mountcastle et al., 90).
More recent behavioral studies in humans using a similar vibrotactile delayed discrimination task indicate that WM for flutter stimuli follows the somatotopic organization of S1 and S2 (Harris et al., 06; Harris et al., 02f; Harris et al., 01). At short delays (< 1 s) subjects were more accurate at comparing vibrations delivered to the same finger than vibrations delivered to the corresponding finger of the opposite hand. Also, accuracy decreased as the distance between the stimulated fingers of one hand increased for shorter (1 s) but not longer (2 s) delays. An interference vibration delivered between the first and the second vibration had the most disruptive effect on accuracy when delivered to the same finger as opposed to different fingers. These results suggest that for shorter delay periods, performance in a somatosensory WM task mirrors the somatotopy and lateralization of processing in S1, whereas at longer delays performance reflects the broader somatotopic and bilateral organization of S2. In addition, disruption of neural activity in S1 using a single pulse of transcranial magnetic stimulation (TMS) impaired performance at short retention intervals (300, 600 ms) but not longer ones (Harris et al., 02). Based on the behavioral effects in their studies, the authors suggested that S1 seems to contribute to human vibrotactile WM for early or short delay periods whereas S2 supports WM for later phases or longer delay periods. However, this interpretation is in contrast to the single-unit studies in non-human primates which did not find delay activity in S1 (Romo und Salinas, 03). Besides possible species differences, these inconsistent results could also be due to differences in experience with the task: whereas the human subjects were unfamiliar with the task before the experiment, the monkeys received extensive training over months probably leading to a more efficient encoding of the vibrotactile stimulus. Recently it has been suggested that these results can also be explained by the adaptation of S1 neurons having an effect on processing in downstream areas involved in early maintenance, possibly S2 (Harris et al., 06; Romo und Salinas, 03). However, early delay activity in S1 was also found in a tactile pattern WM task using single-unit recordings in monkeys (Zhou und Fuster, 96). Thus, the contribution of S1 for the active maintenance of the vibrotactile memory trace during the early part of the delay period remains unclear.
An earlier neuroimaging study on vibrotactile WM using positron emission tomography found higher activity in S2, ventrolateral PFC and PPC during a WM compared to a control condition (Klingberg et al., 96). However, because a continuous discrimination task was employed, activation could not be related to different task periods. Only recently, have the first fMRI studies emerged which used the vibrotactile delayed discrimination task in human subjects (Preuschhof et al., 06; Pleger et al., 06; Li et al., 07; Kostopoulos et al., 07; Burton et al., 07).
Together, the studies in humans suggest 1), that the role of S1 for the early maintenance of vibrotactile stimuli is still vague, and 2), that there might be an additional contribution of PPC but its relative contribution in the different task periods is still unclear. The role of S1 for the active maintenance of the vibrotactile memory trace was further investigated in the present dissertation using neuroscientific methods with adequate temporal resolution: EEG (Study II) and subliminal stimulation (Study III). As mentioned above, additional brain regions associated with vibrotactile WM were identified using fMRI (Study I).
In many studies investigating the discrimination of stimuli varying in quantity (magnitude), the so-called time-order effect (TOE) has been found (Hellstrom, 85; Helson, 64; Fechner, 60). The typical behavioral pattern related to this effect is an interaction between stimulus magnitude and the time-order of stimulus presentation, i.e., whether the stimulus of low or high magnitude was presented at the first (standard stimulus) or at the second (comparison stimulus) position of the trial. For low magnitude stimuli, accuracy increases when the comparison stimulus is of lower magnitude than the standard stimulus. For high magnitude stimuli, accuracy increases when the comparison stimulus is higher than the standard stimulus. Theoretical accounts (Hellstrom, 85) predominantly place the source of the TOE into perceptual and mnemonic processes as opposed to processes taking place during the decision stage of the task (Masin und Fanton, 89d; Masin und Agostini, 91; Masin, 95). Hellström formulated a comprehensive theory that regards the TOE as a side effect of stimulus weighting processes and the effect of a general reference level which is influenced by the stimulus set, context and background information (Hellstrom, 03; Hellstrom, 00; Hellstrom, 85). Similarly, it has been shown that memory for magnitudes is influenced by prior stimuli and determined by regression to the mean of the stimulus set (Huttenlocher et al., 00; Sailor und Antoine, 05). Important insights regarding development of the TOE come from experiments comparing the method of constant stimuli (where an explicit standard and a comparison stimulus have to be compared) and the method of single stimuli (where the comparison has to be judged in relation to the mean of the stimulus set) (Nachmias, 06; Morgan et al., 00). These experiments showed, that independent of the method used, an average standard close to the arithmetic mean of the stimulus set is used by the subjects. Importantly, the generation of this average standard seems to be an implicit, automatic process and only requires the first 10 to 20 trials of the experiment (Morgan et al., 00). Performance in magnitude discrimination, therefore, seems to be influenced by implicit stimulus representations that are based on average information about previous stimulus input. Sinclair and Burton (96) observed the TOE in a vibrotactile delayed discrimination task using a very broad range of standard frequencies (50, 100, 200 Hz) and relatively high differences between standard and comparison frequency. The existence of the TOE indicates that tactile frequency discrimination is influenced by activated LTM representations. However, it remains an open question whether the TOE can also be found when only frequencies within the flutter range are used. In addition, it is not clear how robust this effect is. Therefore, in the present dissertation the TOE was investigated when only flutter frequencies were used. To investigate the robustness of the TOE, task parameters were manipulated (Study IV). Furthermore, the neural correlates of the associated representations remain elusive and until now no study has addressed the neural correlates of the proposed stimulus weighting processes leading to the TOE. To address this issue a parametric fMRI analysis was conducted (Study IV).
|© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.|
|DiML DTD Version 4.0||Zertifizierter Dokumentenserver|
der Humboldt-Universität zu Berlin
|HTML-Version erstellt am: