July 11, 2013 A new study reveals that the representation of complex features in the brain may begin earlier -- and play out in a more cumulative manner -- than previously thought.
The finding represents a new view of how the brain creates internal representations of the visual world. "We are excited to see if this novel view will dominate the wider consensus" said senior author Dr. Miyashita, who is also Professor of Physiology at the University of Tokyo's School of Medicine, "and also about the potential impact of our new computational principle on a wide range of views on human cognitive abilities."
The brain recalls the patterns and objects we observe by developing distinct neuronal representations that go along with them (this is the same way it recalls memories). Scientists have long hypothesized that these neuronal representations emerge in a hierarchical process limited to the same cortical region in which the representations are first processed.
Because the brain perceives and recognizes the external world through these internal images, any new information about the process by which this takes place has the power to inform our understanding of related functions, including knowledge acquisition and memory.
However, studies attempting to uncover the functional hierarchy involved in the cortical process of visual stimuli have tried to characterize this hierarchy by analyzing the activity of single nerve cells, which are not necessarily correlated with neurons nearby, thus leaving these analyses lacking.
In a new study appearing in the 12 July issue of the journal Science, lead author Toshiyuki Hirabayashi and colleagues focus not on single neurons but instead on the relationship between neuron pairs, testing the possibility that the representation of an object in a single brain region emerges in a hierarchically lower brain area.
"I became interested in this work," said Dr. Hirabayashi, "because I was impressed by the elaborate neuronal circuitry in the early visual system, which is well-studied, and I wanted to explore the circuitry underlying higher-order visual processing, which is not yet fully understood."
Hirabayashi and colleagues analyzed nerve cell pairs in cortical areas TE and 36, the latter of which is hierarchically higher, in two adult macaques. After these animals looked at six sets of paired stimuli for several months to learn to associate related objects (a process that can lead to pair-coding neurons in the brain), the researchers recorded neuron responses in areas TE and 36 of both animals as they again performed this task.
The neurons exhibited pair association, but not where the researchers would have thought. "The most surprising result," said senior author Dr. Yasushi Miyashita "was that the neuronal circuit that generated pair-association was found only in area TE, not in area 36." Indeed, based on previous studies, which indicated that the number of pair-coding neurons in area TE is much smaller, the researchers would have expected the opposite.
During their study, Miyashita and other team members observed that in region TE of the macaque cortex, unit 1 neurons (or source neurons) provided input to unit 2 neurons (or target neurons), which -- unlike unit 1 neurons -- responded to both members of a stimulus pair.
"The representations generated in area TE did not reflect a mere random fluctuation of response patterns," explained Dr. Miyashita, "but rather, they emerged as a result of circuit processing inherent to that area of the brain."
In area 36, meanwhile, members of neuron pairs behaved differently; on average, unit 1 as well as unit 2 neurons responded to both members of a stimulus pair. Neurons in area 36 received input from area TE, but only from its unit 2 neurons.
Taken together, these findings lead the authors to hypothesize the existence of a hierarchical relationship between regions TE and 36, in which paired associations first established in the former region are propagated to the latter one. Here, area 36 represents the next level of a so-called feed forward hierarchy.
The work by Hirabayashi and colleagues suggests that the detailed representations of objects commonly observed in the brain are attained not by buildup of representations in a single area, but by emergence of these representations in a hierarchically prior area and their subsequent transfer to the brain region that follows. There, they become sufficiently prevalent for the brain to register. The work also reveals that the brain activity involved in recreating visual stimuli emerges in a hierarchically lower brain area than previously thought.
Moving forward, the Japanese research team has plans to expand upon this research, thus continuing to contribute to studies worldwide that aim to give scientists the best possible tools with which to obtain a dynamic picture of the brain. As a next step, the team hopes to further elucidate interactions between the various cortical microcircuits that operate in memory encoding. Dr. Miyashita has conjectured that these microcircuits are manipulated by a global brain network. Using the results of this latest study, he and colleagues are poised to further evaluate this assumption.
"It will also be important to weave the neuronal circuit mechanisms into a unified framework," said Dr. Hirabayashi," and to examine the effects of learning on these circuit organizations."
Equipped with their new view of cortical processing, the team also hopes to trace the causal chain of memory retrieval across different areas of the cortex. "I am excited by the recent development of genetic tools that will allow us to do this," said Dr. Miyashita. A better understanding of object representations from one area of the brain to the next will shed even greater light on elusive aspects of this hierarchical organ.
Other social bookmarking and sharing tools:
The above story is based on materials provided by American Association for the Advancement of Science, via EurekAlert!, a service of AAAS.
- T. Hirabayashi, D. Takeuchi, K. Tamura, Y. Miyashita. Microcircuits for Hierarchical Elaboration of Object Coding Across Primate Temporal Areas. Science, 2013; 341 (6142): 191 DOI: 10.1126/science.1236927
Note: If no author is given, the source is cited instead.