Annotation of the results obtained in 2021
An analysis of the spatio-temporal structure of EEG signals in the LF and HF ranges was carried out to detect the effects associated with the influence of the top-down and bottom-up information flows on the perception and processing of visual information. The spectral power of the EEG was calculated in the frequency range of 4-40 Hz, which includes the main rhythms of brain activity. The obtained values are compared between the experimental conditions characterizing the processing of the left-oriented and right-oriented Necker cubes on two intervals: TOI1 and TOI2. As a result, differences in spectral power were found in the interval TOI1 in the range of 3-5 Hz on the occipital EEG sensors. The spectral power of the EEG was higher in the case of a right-oriented cube in the case of high ambiguity. The fact that differences were found at TOI1 immediately after the stimulus appeared suggests that they are related to sensory perception and not to high-level decision-making processes (which should dominate at TOI2). This is also confirmed by the localization of the identified differences in the occipital region. The fact that differences were found for complex cubes (but not for simple ones) suggests that they are caused by top-down processes. One of these processes can be selective attention, which allows you to consciously choose the sensory features necessary for making a decision. The fact that increased EEG power is observed for the right-oriented cube indicates that this object is more difficult to perceive, which is confirmed by the results of the response time analysis obtained in the first year of the project. An analysis of the spatio-temporal structure of MEG signals in the LF and HF ranges was carried out to detect the effects associated with the influence of the top-down and bottom-up information flows on the perception and processing of visual information. The spectral power of MEG signals was calculated in the frequency ranges of 4–8 Hz (theta), 8–15 Hz (alpha), 15–30 Hz (beta), 30–40 Hz (gamma), and also for the edge flashing frequencies: 6.7 Hz (left edge blinking) and 8.9 Hz (right edge blinking). The power values obtained are compared between L vs. I and R vs. I condition in all frequency ranges. As a result, differences were found only for the range of 4-8 Hz and for the blinking frequency of the left side. In the frequency range of 4-8 Hz, the spectral power differed on the MEG sensors in the central part of the head, moreover, the sensor areas obtained for L vs I and R vs. I comparisons matched. An analysis of the spectral power on these sensors showed that the spectral power in condition I exceeded the power in conditions L and R. At the blinking frequency of the left side, the spectral power differed on the MEG sensors in the right occipital part of the head, moreover, the sensor areas obtained for L vs I and R vs. I comparisons also matched. An analysis of the spectral power on these sensors showed that the spectral power in condition I exceeded the power in conditions L and R. The result indicates that in condition I, when the influence of the top-down flow is minimal, there is a high spectral power of the MEG in the occipital region at the blinking frequency of the left side, reflecting the dominance of the bottom-up (sensory flows). When the top-down influence begins to dominate (conditions L and R), the spectral power at the flickering frequency drops, indicating a decrease in the bottom-up (sensor) flow influence. It is interesting that the blinking frequency of the right side (8.9 Hz) does not appear on the MEG signals in condition I. This confirms the bottom-up flow asymmetries found in the first year of the project. It should also be noted that the spectral power changes only at a frequency of 4–8 Hz. Moreover, this range contains the blinking frequency and the area of significant differences in the range of 4-8 Hz also overlaps with the area of significant differences at the blinking frequency. It can be assumed that the changes found in the range of 4-8 Hz are caused by the effects described for the blinking frequency. The absence of effects in other ranges can be explained by the fact that the influence of the top-down flow is reflected in integrative processes rather than in segregation processes (local increase in power). To identify them, it is necessary to analyze the functional connections between different areas of the brain in these ranges. An analysis of the functional neural interaction between the top-down and bottom-up flows of sensory information processing based on EEG signals was carried out. As part of solving this problem, matrices of connections between 116 brain regions in the source space were calculated for experimental conditions corresponding to the perception of left-oriented and right-oriented Necker cubes with high and low ambiguity. Connectivity matrices were calculated in the frequency ranges 4-8 Hz (theta), 8-15 Hz (alpha), 15-30 Hz (beta), 30-40 Hz (gamma) for all subjects. As a result of comparing the obtained values, connections were revealed, the strength of which significantly changes when perceiving left-oriented and right-oriented cubes. It should be noted that associations were found both in the case of perception of stimuli with low and high ambiguity for different frequency ranges. An interesting result is that all significant-changing associations are found for the time interval TOI1, which, according to the project hypothesis, contains mainly sensory processing processes. For the TOI2 time interval corresponding to the decision-making processes, no significant differences were found. This confirms the fact that the revealed differences in the structure of functional connections reflect the features of the bottom-up (sensory) flow. Thus, it can be assumed that different features of sensory information (in this case, these are differences in the morphology of left-oriented and right-oriented cubes) are encoded by functional networks with different connection topologies. An analysis of the functional neural interaction between the top-down and bottom-up flows of visual information processing based on MEG signals was carried out. As part of solving this problem, matrices of connections between 116 brain regions in the source space were calculated for the experimental conditions R, L, I described above in this report. By analogy with EEG, connection matrices were calculated in the frequency ranges of 4-8 Hz (theta), 8-15 Hz (alpha), 15-30 Hz (beta), 30-40 Hz (gamma), as well as for the frequencies of face blinking: 6.7 Hz (left side flashing) and 8.9 Hz (right side flashing) for all subjects. As a result of testing the hypothesis that the connectivity strength in condition I exceeds the connectivity strength in conditions R, L, no statistically significant effects were found. On the contrary, when testing the alternative hypothesis, a large number of connections were found in all frequency ranges, for which a statistically significant effect of the change in strength was observed. The results obtained indicate that the process of conscious perception (conditions R, L) leads to an increase in functional interactions in the cortical network of the brain. The data integration process is often seen as a marker of top-down processes, which confirms our theory formulated within the project.

 

Publications

---

Annotation of the results obtained in 2020
We performed experiments to register the electrical activity of the brain (EEG) during the perception and processing of sensory (visual) information. According to the work plan, twenty healthy subjects participated in the experiments. They were all familiar with the experimental paradigm and provided written informed consent. Experimental studies were carried out following the Declaration of Helsinki. An ambiguous two-dimensional drawing of the Necker cube was chosen as a visual stimulus. The subject interprets this two-dimensional image as a three-dimensional left-oriented or right-oriented object. The ambiguity and orientation of a three-dimensional cube depend on the balance between the brightness of the inner edges of a 2D image. By varying the brightness of the inner edges, cubes can be divided into subsets oriented left (LO) and right (RO), as well as images with low ambiguity (LA), which the observer easily interprets, and images with high ambiguity. (HA) images that require more effort to interpret. In the first case, the main influence of the bottom-up flows is expected, and in the second - the top-down flows. We performed experiments to register the magnetic activity of the brain (MEG) during the perception and processing of sensory (visual) information. Following the work plan, 12 conditionally healthy volunteer subjects took part in the experiment. All participants in the experiment provided written informed consent. Experimental studies were carried out under the Declaration of Helsinki. By analogy with the EEG experiment, the visual stimulus was the image of the Necker cube. In this experiment, a stimulus was generated on a computer monitor with a refresh rate of 60 Hz and projected onto a semitransparent screen located in the MEG chamber at a distance of 150 cm from the subject. The brightness of the pixels on the left and right front faces of the cube was modulated with sinusoidal signals with frequencies of 6.67 Hz (60/9) and 8.57 Hz (60/7), respectively. The modulation depth was 100% of the average pixel brightness (127 in 8-bit format), that is, the image brightness varied from black (0) to gray (127). In this experiment, we assumed that the modulation of certain facets would immediately attract the subject's attention under the influence of a bottom-up (sensory) flow. In the case when several faces blink at the same time with different frequencies, attention will switch between them under the influence of internal factors. In the latter case, a greater influence of top-down processes was expected. Within the framework of the project, the average response time of the subject (response time, RT) to visual stimuli was analyzed. It was found that the subjects responded differently to left- and right-oriented stimuli, depending on their ambiguity. When the ambiguity was high, the response times to the left-oriented and right-oriented stimuli were the same. In contrast, with low ambiguity, subjects responded more quickly to stimuli oriented to the left than to stimuli oriented to the right. It can be assumed that the difference in response time between left- and right-oriented stimuli with low ambiguity is the result of differences in the morphology of stimuli. Contrasting EEG wavelet power between left- and right-oriented stimuli in the pre-stimulus period, no significant differences were found. This allows us to conclude that the subject's state under these conditions remained the same (such internal factors as fatigue, attention, and motivation had an equally probable effect on the perception of stimuli under these conditions and could not be the cause of the difference in response time). Contrasting EEG power during stimulus processing, a negative cluster was found with a significance level of p = 0.042 in the 3-4 Hz frequency band. This cluster formed from 0.23 s to 0.5 s after the appearance of the stimulus and included EEG sensors in the right-sided occipital and parietal regions. The average EEG power in this cluster was higher when the subject processed stimuli oriented to the right than stimuli oriented to the left. According to [Okazaki et al, // Neurosci. research (2008)], morphology for an ambiguous stimulus does not differ between different interpretations. In other words, the bottom-up process associated with sensory data does not carry information that allows the stimulus to be interpreted unambiguously. In this case, the decision is made under the influence of the top-down mechanisms such as memory, experience, and contextual templates [Engel and Fries // Curr. opinion neurobiology (2010). Wang, Arteaga, and He // Proc. Natl. Acad. Sci. (2013)]. Thus, an increase in the ambiguity of the Necker cube increases the influence of top-down factors. When the ambiguity is small, left- and right-oriented Necker cubes have different morphologies. As the ambiguity increases, all interior edges become equally prominent and the morphology of the left- and right-oriented cube begins to coincide. In this context, the result obtained can be interpreted as follows. With a low ambiguity, the subjects responded more quickly to stimuli of the left orientation. On the contrary, for high ambiguity, the response time to stimuli oriented to the left and to the right does not practically differ. Consequently, the observed effect is most likely caused by the peculiarities of the bottom-up mechanisms and decreases when the top-down processes dominate. In the post-stimulus period, the EEG power for left- and right-oriented stimuli with low ambiguity differed. It can be assumed that this reflects different bottom-up processing mechanisms caused by morphological differences. No changes in EEG power were observed for stimuli with high ambiguity and similar morphology. The main change in EEG power was concentrated in the right hemisphere, at the temporoparietal junction (TPJ), which is part of the ventral attention network (VAN). Analysis of the scientific literature indicates that VAN is lateralized to the right hemisphere and controls the bottom-up information processing [Farrant and Uddin // Dev. cognitive neuroscience (2015)]. Thus, activation of the right TPJ after the stimulus onset may indicate an increase in the influence of the bottom-up mechanism. It was shown that the EEG power in the right TPJ increases for both left-oriented and right-oriented stimuli, reflecting the processing of their morphological features. Stimuli oriented to the right induce higher power in the right TPJ. These results suggest that the right-oriented stimulus has morphological features that require the involvement of large bottom-up resources to process them. In the MEG experiments, no behavioral response was recorded, which makes it possible to understand what interpretation of the stimulus the subject chose. Instead, the morphological features of the cube corresponding to the left and right orientations were labeled in a certain way - their brightness was modulated by a harmonic law with the corresponding frequencies f1 and f2. These frequencies were detected in signals of brain activity. At the first stage, the coefficient D was analyzed, which characterizes the difference in the power of the wavelet spectrum of the MEG at the corresponding frequencies. In the case when the modulation was applied to the features of the left-oriented cube, the D value took positive values in the majority of the subjects in the group. In the case when modulation was applied to the features of a right-oriented cube, D took negative values. Thus, the modulation made it possible to statistically reliably detect the moments of focusing of the subjects on the corresponding morphological signs of the stimulus. Within the framework of the project, we developed methods for the analysis of functional neural interaction in a certain frequency range and between high-frequency and low-frequency ranges. The functional connection between neural ensembles was assessed on the basis of the phase coherence of their activity in a certain frequency range. At the first stage, using EEG signals, the phases of activity of the neural ensembles of the internal parts of the brain were restored. Based on the obtained data obtained, we calculated matrices containing the coefficients of phase coherence between neural ensembles of different anatomical zones of the brain. These matrices were calculated for a group of subjects, for stimuli with high and low ambiguity in different frequency ranges: theta (4-7 Hz), alpha (8-12 Hz), beta (13-30 Hz), and gamma (30-46 Hz). As a result of a statistical comparison of the obtained matrices between stimuli with high and low ambiguity, significant differences were found only in the ranges of 8-12 Hz and 13-30 Hz. It was found that the largest number of changing functional connections are concentrated in the range of 13-30 Hz. The leading role of this frequency range in the processing of sensory ambiguity is also noted in the works of other authors [Okazaki et al. // Neurosci. research (2008). Engel and Fries // Curr. opinion neurobiology (2010)], which associates this frequency range with the activation of top-down processes. The results obtained indicate a large number of functional connections emanating from the central region of the brain to the periphery. According to the anatomical atlas, this central point marks the region of the anterior cingulate cortex, which is the source of the top-down processes [Chen et al. // Nature communications (2018)]. A phase-locking value (PLV) was used to assess neural interactions between different frequency ranges. Using this characteristic, the indices of phase coherence were calculated between signals of neural activity when processing stimuli with high and low ambiguity in different frequency ranges: theta-alpha, alpha-beta, beta-theta. Then, the obtained values were averaged over all EEG channels. The mean values of the coherence index, in turn, were compared using analysis of variance ANOVA. As a result, a significant effect of ambiguity and frequency bands was found. It was shown that an increase in stimulus ambiguity leads to an increase in the phase coherence index between different frequency ranges. In this case, the coherence index turns out to be the higher, the closer to each other the analyzed ranges are located and the higher their frequencies. This result is explained by the fact that high-frequency rhythms support the communication of neurons in small local ensembles. Low-frequency rhythms, in turn, carry out functional interactions between neural ensembles located far from each other. The fact that the coefficient of coherence is higher for higher frequency ranges suggests that the activity of neurons (and their functional interaction) is higher within local areas. That is, the neural network demonstrates the property of functional segregation. At the same time, with increasing stimulus ambiguity, the phase coherence index increases both between high-frequency and low-frequency rhythms. This means that in addition to functional segregation, an increase in functional integration is also observed (increased interaction between distant brain regions). The latter is also a sign of top-down processes of brain activity. Summarizing the above, we developed experimental paradigms based on ambiguous visual stimuli (Necker cubes), which made it possible to separate the influence of the bottom-up and top-down information processing processes by manipulating the degree of ambiguity and orientation of the stimulus. Within the framework of these paradigms, we performed experimental studies to record EEG, MEG signals, and behavioral responses. A preliminary analysis of the data was carried out and various experimental conditions were identified. Finally, we developed methods to analyze functional neural interactions in a specific frequency range and between high-frequency and low-frequency ranges. These methods have been successfully tested on the obtained data. The results obtained in the first year of the project are the basis for solving the tasks planned in the second year.

 

Publications

---