Visual processing

Local-feature assembling in visual pattern recognition and generalization in honeybees. Stach, S. et al. Nature 429, 758–761 (2004)

Continuing on a theme of recent years in demonstrating increasingly complex sensory and cognitive abilities in the tiny brains of bees, the authors show that honeybees can link different features in learning a visual pattern, and can generalize their responses to novel stimuli that share the layout of the trained stimuli. Generalization depended on the number of edges that were the same between two layouts. The pattern recognition required stimulation of the achromatic L-photoreceptor system.

Development

In vivo convergence of BMP and MAPK signaling pathways: impact of differential Smad1 phosphorylation on development and homeostasis. Aubin, J. et al. Genes Dev. 18, 1482–1494 (2004)

Effector molecules called Smads integrate signalling pathways that are important during development and in homeostasis. Aubin et al. show that a carboxy-terminal mutant Smad that cannot undergo transcriptional activation by BMP (bone morphogenetic protein) recapitulates many of the features of a Smad-null mutant, whereas a mutant in which Smad cannot be phosphorylated by MAPK (mitogen-activated protein kinase) has various other phenotypes. It seems that signalling by BMP and MAPK through Smads needs to be precisely balanced.

Neurotechniques

Localized chemical release from an artificial synapse chip. Peterman, M. C. et al. Proc. Natl Acad. Sci. USA 101, 9951–9954 (2004)

The authors have developed a chip that uses electro-osmosis to deliver tiny quantities of chemical compounds through 5-μm apertures. They cultured PC12 cells on this 'artificial synapse chip' and can stimulate individual cells by controlled release of bradykinin. As a prototype neural interface, the chip could have many applications in basic research and clinical treatments.

Cognition

Matching behavior and the representation of value in the parietal cortex. Sugrue, L. P. et al. Science 304, 1782–1787 (2004)

Both reward history and reward probability influence decision making. To investigate how these factors are represented by neurons, Sugrue et al. used an eye-movement task in which monkeys were rewarded with different probability for choosing different targets. The behaviour of the monkeys tracked changing reward probabilities and could be predicted by a simple model. The authors found that neurons in the parietal cortex represented the relative value of available choices as predicted by the model.