The Neural Mechanism Of Spontaneous Numerical Processing

The competence of numerical cognition is one of higher cognitive competences possessed by human and many non-human animals. Evolution endowed our brain the compressed representation of quantity to adapt the rigorous survival and living. The compressed representation was carried out as Dehaene and Changeux indicated that numerical information was extracted from visual scenes, and was processed sequentially by summation coding and place coding. At the stage of summation coding, neurons discharged in a linear relationship with the amount of numbers. And at the stage of place coding, neurons discharged in a logarithmic relationship with numbers, in accordance with the Weber-Fechner law.Many studies confirmed that the place coding existed in the intraparietal sulcus in rhesus macaques and human beings, and in the prefrontal cortex of monkeys. But only one study conducted by Roitman et al. found that summation neurons located in lateral intraparietal sulcus. Evidence from neuroimaging studies revealed that human intraparietal sulcus also showed some attributes of place coding. Besides, these two coding modes were mainly found at the range of small numbers, and small numbers and large numbers was considered to be processed distinctly, so some researchers questioned whether these coding modes were common for all numbers or they were effective only at the range of small numbers. The current study focused on two main aspects:first, whether summation coding and place coding could be executed sequentially, and second, whether they were common across ranges of small numbers and large ones.Most previous studies which found numerical effects in the intraparietal sulcus adopted various comparison paradigms. Meanwhile, other studies about the comparison of continuous quantities also found the activity of intraparietal sulcus. These methods confused the numerical representation and the magnitude comparison. On the other hand. Dr. Verguts suggested that coding strategies could be specific to the task used. In a same-different judgment task, place coding might be a more efficient way of coding for this specific task; contrarily, in a small-large judgment task, summation coding might be a more efficient way of coding for that task. The difficulty and also the originality of the present study was that segregating pure processing of numerical cognition with the influence of reaction and magnitude comparison via an elegant and ingenious experiment.In the first experiment, the natural separation between subitizing and counting was examined by reaction time technique. The behavior data showed that reaction time grew like a power function in pace with the increase of number, the growing was immune to the very short presentation time. Beyond a higher ecological validity, the experiment revealed the nonlinear mode of growing of counting time for the first time.In the second experiment, the pure processing course of numerical cognition without movement response was investigated by adopting the Irrelevant Task Paradigm (ITP). In irrelevant task paradigm, stimuli arrays were presented similar to the oddball paradigm, only the deviant stimulus was required to response. Theoretically, the standard stimulus and the deviant stimulus changed in different dimensions. This creative paradigm ruled out the impact of specific task and response preparation. This paradigm effectively avoided the influence of probability and interaction among different factors. Considerable progress has been made in two aspects:Firstly, the difference between subitizing and counting seem to be at the earlier stage of processing course. The P1 component showed that numbers in the range of counting were markedly different from numbers in the range of subitizing. This finding clarified previous debates that whether subitizing was pre-attentive or attentive dependent, and implied that it maintained a close relationship with visuo-spatial attention and working memory.Secondly, both summation coding and place coding were showed to reside in the course of spontaneous numerical processing, and were not task dependent. The two coding modes were executed sequentially, and reflected by N1 component and P2p component, respectively. These results were in line with the neural network model brought forward by Dehaene and Changeux.In a word, the current study powerfully revealed the neural mechanism of numerical processing, by adopting a creative paradigm. But there are still some questions need to be investigated in future studies, such as the role of visuo-spatial attention and working memory in numerical processing, the extraction of numerical information from visual scenes, the universality of these two coding modes in numerical processing, the physiological process of different neural coding, and so on.