The cognitive divide assumed to separate humans and other animals continues to get narrower. It has, for example, recently been discovered that elephants have self awareness, and that African cichlids use simple logic to infer their position in the social hierarchy; earlier this year, a research team from the University of Cambridge showed that scrub jays plan for the future, while anthropologists from the University of Iowa observed chimps making spears and using them to stab bushbabies; and a paper by researchers from the University of Georgia, published earlier this month, suggested that rats are capable of metacognition (the ability to assess their own knowledge).
Now, there is evidence that mice can form abstract concepts, an ability which was also believed to be exclusive to humans and, possibly, non-human primates. Joe Tsien and his colleagues at Boston University’s Center for Systems Neurobiology, together with a group of collaborators in China, report the identification of “nest cells” in the mouse brain. The cells, found in the CA1 region of the hippocampus, exhibit dynamic changes in activity when the mice encounter an appropriately sized compartment which can be used as a nest. The findings are published online today, in the Proceedings of the National Academy of Sciences.
Here’s the abstract:
As important as memory is to our daily functions, the ability to extract fundamental features and commonalities from various episodic experiences and to then generalize them into abstract concepts is even more crucial for both humans and animals to adapt to novel and complex situations. Here, we report the neural correlates of the abstract concept of nests or beds in mice. Specifically, we find hippocampal neurons that selectively fire or cease to fire when the mouse perceives nests or beds, regardless of their locations and environments. Parametric analyses show that responses of nest cells remain invariant over changes in the nests’ physical shape, style, color, odor, or construction materials; rather, their responses are driven by conscious awareness and physical determination of the categorical features that would functionally define nests. Such functionality-based abstraction and generalization of conceptual knowledge, emerging from episodic experiences, suggests that the hippocampus is an intrinsic part of the hierarchical structure for generating concepts and knowledge in the brain.
The authors used multiple microelectrode ensembles, embedded in the animals’ hippocampi, to simultaneously record the electrical activity of groups of about 200 hippocampal cells while the mice roamed around their cages. Three types of nest cell were identified, each with a different firing pattern. One type, named the transient-on type, drastically increased its firing, from a basal rate of approximately 1.5 Hz to about 24 Hz, whenever the animals approached the nest or a similar concave object. Another, called the persistent-on cell, exhibited an even more dramatic increase in firing: the basal firing rate of 0.3 Hz increased to up to 40 Hz whenever the animals entered the nest, and remained at the elevated state of activity as long as the animals stayed inside. The third type, named the persistent-off cell, had a firing pattern inverse to that of the persistent-on cells; that is, their activity decreased when the mice were in the nests.
This film clip shows a mouse climbing into a “nest”; the trace at the bottom of the screen shows the firing patterns of transient-on cells, whose activity increases when the mouse enters the nest:
The firing patterns of the nest cells are similar to place cells, which fire when the animal is in a specific location, and encode episodic memories (that is, memories relating to past experiences). There are far fewer nest cells than place cells in the mouse hippocampus. Unlike place cells, their activity increases regardless of location – the same responses were recorded when the make-shift nests, consisting of various types of small containers, were moved around the cages in which the experiments were carried out. They fired in response to specific, fundamental features extracted from the environment, regardless of the context in which those features are encountered. Nest cells therefore encode semantic memories, which are related to the meaning of an object. They respond to the functional properties of the object – if an appropriately-sized container was turned upside-down or covered with a piece of glass, no response was recorded from them, but as long as a container can serve as a nest, the cells increase or decrease their activity.
On the basis of their findings, the authors suggest that nest cells are involved in encoding the abstract representation of the “nest” in the mouse brain. This study also provides evidence of hippocampal processing units involved in extracting abstract features from objects in the environment. These units, called neural cliques, consist of groups of cells with similar selectivity and response properties. Some neural cliques are highly selective, responding only to specific shocking stimuli, such as shaking movements, drops, or puffs of air blown onto the animal (the earthquake-specific, drop-specific, and air puff-specific neural cliques, respectively). Other cliques are less selective: cells in the general startle clique respond to any startling stimulus, while cells in the sub-general startle clique fire when a specific combination of two startling events is encountered.
In the visual cortex, different classes of cells respond to specific properties of objects, such as colours, corners or the orientiation of edges. At successively higher levels of processing, carried out in the association cortices, these properties are combined so that the representation of the object becomes increasingly complex. Hippocampal networks consisting of neural cliques acting together encode internal, abstract representations of external objects. Exactly how is unclear, but they probably act in a categorical, combinatorial and hierarchical manner, like cells in the visual cortex.
(Thanks to Roxanne Khamsi for drawing my attention to this paper. Roxanne is the first person to contact me via the blog’s new email address, which I added to the About page earlier this week. Read her summary of the study at New Scientist.)
Lin, L., et al. (2007). Neural encoding of the concept of nest in the mouse brain. PNAS doi: 10.1073/pnas.0701106104 [Abstract]
Lin, L., et al. (2006). Organizing principles of real-time memory encoding: neural clique assemblies and universal neural codes. Trends Neurosci. 29: 48-57. [Full text]