The paper analyzes the work of Russian and foreign scholars devoted to the role of cross analyzer cooperation in developing and implementing mathematical abilities. Crossmodal interaction is considered as an additional category of neuropsychological analysis that allows to extend the existing ideas about the psychological structure and brain providing the mathematical ability. There are data that confirm the relevance of studying the interaction of the senses. Many of the research on this issue are carried out using the synesthesia which is considered a rare phenomenon. However, both Russian and foreign works suggest that the interaction of analyzers is not characteristic only to those whose brain is synesthetic. The joint work of the senses is characteristic of every person since his/her childhood, and is an obligatory condition for cognitive processes.

Cross analyzer synthesis is assumed to play an important role in producing spatial representations and the ability to intuitively perceive the notion of quantity (evolutionary foundations of mathematical ability). On the brain level, these processes are provided primarily by functioning of parietal and tertiary cortical areas located at the junction of cortical analyzer areas and also temporal areas that border on the parahippocampal brain area.

When dealing with school mathematics the structure of mathematical abilities is changing due to verbal and symbolic representations of numerical coding. Dealing with symbols opens up new opportunities, but it also narrows the spectrum of modalities involved in doing mathematical sums. Thus, the ability to re-encode information from one modality to another after school mathematics is perceived has an impact on the efficacy of mathematical activity. Doing mathematical sums is accompanied by crossmodal interaction that occurs on the unconscious level.

Some problem conditions may be efficiently processed in one modality, others may be solved in other modality.

Apparently, the ability to various crossmodal re-encoding patterns varies considerably from person to person. The effectiveness of crossmodal interactions may determine the severity of certain components of mathematical abilities and influence successful solutions of the corresponding types of mathematical problems.

The paper studies the connection of the lateral features in the manual, auditory, visual fields, and their interaction with the mathematical abilities in adolescence. The research involved 92 healthy people including 17 males and 75 females aged 15 to 25 years (18,7 ± 2,2), who do not major in mathematics. To measure the components of mathematical ability the standardized math test MAAGS-2015 to reveal arithmetic, algebraic, geometric abilities was used. Diagnosis of hemispherical asymmetry was performed using self-reports of manual asymmetry, M.Annette standardized questionnaire modification, samples of «Interlocking fingers,» «Napoleon’s Pose», «Applause», «Aiming», Rosenbach’s test and dichotic listening. When comparing the features with lateral components of mathematical ability to incorporate asymmetries possible interaction of different analyzers was considered.

The research results reveal that different lateral characteristics are significant predictors for the different components of mathematical abilities; some lateral symptoms are not related to mathematical ability. The greatest predictive power belongs to sensory asymmetries and their interaction. In general, the highest mathematical abilities are observed in patients with right and bilateral signs, left-sided symptoms often reveal negative predictors. The interaction asymmetries between different analyzers manifested in unequal due to the mathematical abilities indicators lateralization in the same field in different versions of lateralization in the other. Cross-lateralization in most cases is a negative predictor of mathematical abilities. The models based on the interaction between the lateral features allow to explain more than a quarter of the variability of the components of mathematical abilities. The predictive ability of these models is significantly higher than that of models with individual predictors.