1. CURSOS DE SECUNDARIA. 2º- 3º- 4º MUSICA
1.3.2. INSTRUMENTOS DE EVALUACIÓN
Studies with non-human primates revealed sustained neural response during the maintenance interval across many neurons in the inferotemporal cortex [Fuster and Jervey,1982,Miller et al.,1993]. An analogous sustained brain response has been observed in the human EEG recorded over the posterior parietal and occip-ital sites. It has been called the contralateral delay activity (CDA) because it is observed during the delay period on the side contralateral to the location of memorized material [Vogel and Machizawa,2004,Vogel et al.,2005]. In particular
it is measured at posterior parietal, lateral occipital and posterior temporal elec-trode sites contralateral to the location of memorized stimuli. This suggests that an underlying neural mechanism is anatomically unspecific and rather a global phenomenon observed across many brain networks. Vogel and Machizawa [2004]
observed that the amplitude of the CDA increases with increasing working mem-ory set size. Importantly, the amplitude rises only up to about four items and saturates around an individual WM limit. The authors observed that individual WM capacity limit was strongly positively correlated with the increase of CDA’s amplitude from memory load 2 to memory load 4. Vogel et al. [2005] further observed that the individual limit in the number of items memorized in WM (in-dexed by the CDA), strongly depends on the efficiency in selecting the relevant information. In fact, participants with high and low WM capacity were able to maintain about the same number of items. However, the former group of subjects was more efficient in selecting the relevant items (i.e. ignoring irrelevant distrac-tors). In contrast, the low-capacity participants although capable to memorize a similar number of items were more likely to maintain both relevant and irrelevant items.
Another model describing the limits in WM capacity has been proposed byLisman and Idiart [1995]. Unlike Vogel’s approach it relies on the dynamic interactions between theta and gamma oscillations (see Figure 2.3A). The authors suggested that information might be sustained by increased membrane excitability repeated on each cycle of a network oscillating with the theta (5-12Hz) frequency. Accord-ing to the model each memory item is stored as a neural assembly firAccord-ing within the gamma cycle whereas the whole memory set is refreshed with the theta-alpha oscillation (5-12Hz). The model is able to explain several observations reported by the psychophysics experiments. For example, it explains the limit in WM ca-pacity to about 7 ± 2 items as observed by Miller (see chapter 1.4). According to the model this limit reflects the ratio between gamma activity (∼40Hz; repre-senting individual items) and theta-alpha oscillations (5-12Hz; refreshing memory set). This ratio between frequencies imposes a limit in the number of items that might be refreshed in a single low-frequency cycle. The items represented by cell
alpha code
B
CFC strength depends on duration of duty cycle
item 1
item 2
theta-gamma code gamma signal duty cycle
A item 1 item 2
9 5 2 83 9 5 2 8 3
Figure 2.3: The multiplexing model of working memory. (A) The initial multiplexing model developed by Lisman and Idiart [Lisman and Idiart,1995].
The model suggests a coupling between the phase of low-frequency oscillations in theta and alpha (5-12Hz) range and the gamma oscillations (∼40Hz). Each item is represented as a neural assembly synchronized in the gamma cycle.
Consecutive gamma cycles represent corresponding items in a memory sequence.
The model further suggests that the whole memory set is reactivated during successive theta cycles. (B) Suggested extension of the multiplexing model.
The strength of CFC is not uniform during maintenance and depends on the length of the alpha duty cycle. The length of the duty cycle relies on power in the alpha frequency range. Low alpha power (orange on the left) results in longer duty cycles, whereas high alpha power (orange on the right) is associated with shorter duty cycle. The length of the duty cycle constrains gamma power (indicated by green oscillations), which in turn may be reflected as modulations to the theta-gamma code. In particular, long duty cycles may be observed as decreases in strength of the theta-gamma modulation index (modulating activity is broadly distributed across the low-frequency phase; upper left). In contrast, short duty cycles may be reflected by increases in CFC strength (upper right).
assemblies firing after the preferred window of membrane excitability will not be reactivated on consecutive theta-alpha cycles and therefore will fade away. Thus, the model predicts the inter-individual correlation between WM capacity and the oscillatory signal properties. In particular, the longer the preferred window of membrane excitability (indexed by the frequency of individual theta dominant oscillation), the higher individual WM capacity. Alternatively, one might also expect that the faster synchronization of neural assemblies representing individ-ual items (indexed by the dominant frequency of gamma oscillations), the higher individual WM capacity. Furthermore, the model explains the results of Stern-berg experiments on the maintenance of multiple items in WM (see chapter 1.5).
If representation of each item is stored by cells firing in the gamma cycle and individual items are reflected by firing at different phases of the theta-alpha oscil-lations then larger memory set size increases the number of cell assemblies locked to the consecutive theta-alpha phases. In other words, larger memory load is as-sociated with increased portion of the slow oscillation which already is asas-sociated with a gamma activity representing individual items. This chain of representations increases exactly by the duration of gamma cycle with each item in the memory set. This explains Sternberg’s observation that the reaction time increases linearly at about 38ms per memory item. Note that this explanation requires an addi-tional assumption that items are recalled serially one-by-one which indeed was proposed by Sternberg. The multiplexing model of WM offers several clear-cut predictions. The number of gamma sub-cycles within the theta cycle during WM maintenance should scale with number of items maintained. The distribution of gamma activity across theta cycle should be non-uniform, resulting in the cross-frequency coupling (CFC). Indeed, several studies observed increased CFC during WM maintenance [Axmacher et al., 2010b, Chaieb et al., 2015,Leszczy´nski et al., 2015, Sauseng et al., 2009, Siegel et al., 2009] corroborating the concept of the CFC as a potential mechanism for storing multiple items in WM.