A random-matrix theory of the number sense

Spontaneous supra-modal encoding of number in the infant brain

The core knowledge hypothesis postulates that infants automatically analyze their environment along abstract dimensions, including numbers. According to this view, approximate numbers should be encoded quickly, pre-attentively, and in a supra-modal manner by the infant brain. Here, we directly tested this idea by submitting the neural responses of sleeping 3-month-old infants, measured with high-density electroencephalography (EEG), to decoders designed to disentangle numerical and non-numerical information. The results show the emergence, in approximately 400 ms, of a decodable number representation, independent of physical parameters, that separates auditory sequences of 4 vs. 12 tones and generalizes to visual arrays of 4 vs. 12 objects. Thus, the infant brain contains a number code that transcends sensory modality, sequential or simultaneous presentation, and arousal state.

Role of Lateral Inhibition on Visual Number Sense

Newborn animals, such as 4-month-old infants, 4-day-old chicks, and 1-day-old guppies, exhibit sensitivity to an approximate number of items in the visual array. These findings are often interpreted as evidence for an innate “number sense.” However, number sense is typically investigated using explicit behavioral tasks, which require a form of calibration (e.g., habituation or reward-based training) in experimental studies. Therefore, the generation of number sense may be the result of calibration. We built a number-sense neural network model on the basis of lateral inhibition to explore whether animals demonstrate an innate “number sense” and determine important factors affecting this competence. The proposed model can reproduce size and distance effects of output responses of number-selective neurons when network connection weights are set randomly without an adjustment. Results showed that number sense can be produced under the influence of lateral inhibition, which is one of the fundamental mechanisms of the nervous system, and independent of learning.

Extraordinary claims, extraordinary evidence? A discussion

Roberts (2020, Learning & Behavior, 48[2], 191-192) discussed research claiming honeybees can do arithmetic. Some readers of this research might regard such claims as unlikely. The present authors used this example as a basis for a debate on the criterion that ought to be used for publication of results or conclusions that could be viewed as unlikely by a significant number of readers, editors, or reviewers.

Numerosity representation in a deep convolutional neural network

Enumerating objects in the environment (i.e., “number sense”) is crucial for survival in many animal species, and foundational for the construction of more abstract and complex mathematical knowledge in humans. Perhaps surprisingly, deep convolutional neural networks (DCNNs) spontaneously emerge a similar number sense even without any explicit training for numerosity estimation. However, little is known about how the number sense emerges, and the extent to which it is comparable with human number sense. Here, we examined whether the numerosity underestimation effect, a phenomenon indicating that numerosity perception acts upon the perceptual number rather than the physical number, can be observed in DCNNs. In a typical DCNN, AlexNet, we found that number-selective units at late layers operated on the perceptual number, like humans do. More importantly, this perceptual number sense did not emerge abruptly, rather developed progressively along the hierarchy in the DCNN, shifting from the physical number sense at early layers to perceptual number sense at late layers. Our finding hence provides important implications for the neural implementation of number sense in the human brain and advocates future research to determine whether the representation of numerosity also develops gradually along the human visual stream from physical number to perceptual number.

Visual number sense in untrained deep neural networks

Number sense, the ability to estimate numerosity, is observed in naïve animals, but how this cognitive function emerges in the brain remains unclear. Here, using an artificial deep neural network that models the ventral visual stream of the brain, we show that number-selective neurons can arise spontaneously, even in the complete absence of learning. We also show that the responses of these neurons can induce the abstract number sense, the ability to discriminate numerosity independent of low-level visual cues. We found number tuning in a randomly initialized network originating from a combination of monotonically decreasing and increasing neuronal activities, which emerges spontaneously from the statistical properties of bottom-up projections. We confirmed that the responses of these number-selective neurons show the single- and multineuron characteristics observed in the brain and enable the network to perform number comparison tasks. These findings provide insight into the origin of innate cognitive functions.

Numerosity discrimination in deep neural networks: Initial competence, developmental refinement and experience statistics

Both humans and non-human animals exhibit sensitivity to the approximate number of items in a visual array, as indexed by their performance in numerosity discrimination tasks, and even neonates can detect changes in numerosity. These findings are often interpreted as evidence for an innate 'number sense'. However, recent simulation work has challenged this view by showing that human-like sensitivity to numerosity can emerge in deep neural networks that build an internal model of the sensory data. This emergentist perspective posits a central role for experience in shaping our number sense and might explain why numerical acuity progressively increases over the course of development. Here we substantiate this hypothesis by introducing a progressive unsupervised deep learning algorithm, which allows us to model the development of numerical acuity through experience. We also investigate how the statistical distribution of numerical and non-numerical features in natural environments affects the emergence of numerosity representations in the computational model. Our simulations show that deep networks can exhibit numerosity sensitivity prior to any training, as well as a progressive developmental refinement that is modulated by the statistical structure of the learning environment. To validate our simulations, we offer a refinement to the quantitative characterization of the developmental patterns observed in human children. Overall, our findings suggest that it may not be necessary to assume that animals are endowed with a dedicated system for processing numerosity, since domain-general learning mechanisms can capture key characteristics others have attributed to an evolutionarily specialized number system.

Overexpression of phosphoprotein enriched in astrocytes 15 reverses the damage induced by propofol in hippocampal neurons

Propofol is a general anesthetic used in surgical operations. Phosphoprotein enriched in astrocytes 15(PEA15) was initially identified in astrocytes. The present study examined the role of PEA15 in the damage induced by propofol in hippocampal neurons. A model of hippocampal neuron damage was established using 50 µmol/l propofol. Cell viability, proliferation and apoptosis of hippocampal neurons were tested by Cell Counting Kit‑8 and flow cytometry. Western blotting and reverse transcription‑quantitative polymerase chain reaction analysis were performed to measure the expression levels of PEA15, and additional factors involved in apoptosis or in the signaling pathway downstream of PEA15. The present results suggested that propofol significantly decreased PEA15 expression levels in hippocampal neurons. Furthermore, overexpression of PEA15 significantly increased the cell viability and cell proliferation of cells treated with propofol. Additionally, PEA15 overexpression decreased apoptosis, which was promoted by propofol. Treatment with propofol significantly decreased the protein expression levels of pro‑caspase‑3, B‑cell lymphoma-2, phosphorylated extracellular signal‑regulated kinases (ERK)1/2, ribosomal S6 kinase 2 (RSK2) and phosphorylated cAMP responsive element binding protein 1 (CREB1). However, propofol upregulated active caspase‑3 and Bax expression levels. Notably, PEA15 overexpression was able to reverse the effects of propofol. Collectively, overexpression of PEA15 was able to attenuate the neurotoxicity of propofol in rat hippocampal neurons by increasing proliferation and repressing apoptosis via upregulation of the ERK‑CREB‑RSK2 signaling pathway.

Decay of connection probabilities with distance in 2D and 3D neuronal networks

We study connectivity in a model of a growing neuronal network in dimensions 2 and 3. Although the axon-to-dendrite proximity is an insufficient condition for establishing a functional synapse, it is still a necessary one. Therefore we study connection probabilities at short distances between the randomly grown axon trees and somas as probabilities of potential connections between the corresponding neurons. Our results show that, contrary to a common belief, these probabilities do not necessarily decay polynomially or exponentially in distance, but there are regimes of parameter values when the probability of proximity is not sensitive to the distance. In particular, in 3 dimensions the Euclidean distance between the neuronal cell body centers of neurons seems to play a very subtle role, as the probabilities of connections are practically constant within a certain finite range of distance. The model has a sufficient number of parameters to assess networks of neurons of different types. Our results give a firm basis for further modelling of the neuronal connectivity taking into account some realistic bouton distributions for establishing synaptic connections.

Different reaction-times for subitizing, estimation, and texture

Humans can estimate and encode numerosity over a large range, from very few items to several hundreds. Two distinct mechanisms have been proposed: subitizing, for numbers up to four and estimation for larger numerosities. We have recently extended this idea by suggesting that for very densely packed arrays, when items are less segregable, a third "texture" mechanism comes into play. In this study, we provide further evidence for the existence of a third regime for numerosity. Reaction times were very low in the subitizing range, rising rapidly for numerosities greater than four. However, for tightly packed displays of very high numerosities, reaction times became faster. These results reinforce the idea of three regimes in the processing of numerosity, subitizing, estimation, and texture.