246 cuss receptor biophysics; Victor discusses non-linearities in the cat retina; Hubel and Cowan discuss the anatomy and development of the monkey visual pathways: Wilson and Julesz review their own work on spatial frequency channels and texture discrimination, while Mart and Ullman describe their own computational theories of early visual processing. A final chapter by Bizzie describes an interesting experiment on visuo-motor control in the monkey. The central and most original theme of the book, which gives unity to much of this apparently disparate subject matter, is its treatment of non-linear transformations. All organisms are non-linear in the way on which they respond to stimuli, while even at the earliest stages of neural processing, fundamentally non-linear interactions occur. While the mathematical analysis of linear systems is straightforward, that of non-linear systems is not. Traditionally. Wiener analysis has been regarded as the most powerful and general way to study non-linearities, but as is pointed out in the
book. there are mathematical and practical difficulties associated with its use. Furthermore, in Poggio's words "the computational properties of a system will usually remain hidden in the numerical values of the kernals nearly as much as in the original input-output data'. As an alternative to this, Poggio and Reichardt show that an analysis in terms of a Volterra-like series is often valid for biological systems. They develop a quasi-graphical notation for the term,s in the series and show how different types of graph can characterize many different types of computation, such as the detection of velocity, relative velocity, or the second order statistics of the textures studied by Julcsz. More speculatively, Poggio and Tone propose that any particular graphical representation may have a natural interpretation in terms of neural circuitry in which computations are performed by non-linear local circuit interactions between strategically placed excitatory and inhibitory synapses. As a non-mathematician, I l:bund this analysis hard to understand, and 1 was not
M o n o c l o n a l s - the story so far M o n o c l o n a l A n t i b o d i e s to N e u r a l Antigens edited by Ronald McKay, Martin C. Raff and Louis F. Reichardt, Cold Spring Harbor Laboratory, 1981. $35.00 U.S./$42. O0 elsewhere (xv + 282 pages) ISBN 0 87969 138 7
Like many of my fellow book reviewers in these columns, I have developed an acute immune reaction (manifesting primarily as a foaming at the mouth) in response to reviewing publications that comprise camera-ready conference papers, which masquerade under seductive titles as books. This 'tertiary literature' usually comprises unrefereed, unedited reiterations of material that has or will be published elsewhere. Those who participate in its genesis do so reluctantly because it has become normal practice for organizers of symposia to demand a manuscript as a condition of participation. Monoclonal Antibodies to Neural Antigens. containing the proceedings of a meeting held at Cold Spring Harbor, New York in late 1980, escapes many of these strictures. The book comprises 26 brief but generously illustrated chapters grouped into 5 sections - (1) Defining Neuronal Cell Types and Cell Lines; (2) Defined Antigens; (3) The Synapse; (4) The Retina; (5) The Neuromuscular Junction. Each manuscript was transcribed from the talks pre-
sented by a set of 'scribes' and then edited, first by the presenter and then by the book editors. Little purpose would be served in attempting to review each or even some of the individual chapters. Two primary advances appear to have been made by the use of monoclonal antibodies thus far in neurobiotogy. The first is the production of specific antisera against known immunogens that are difficult to purify to homogeneity (unfortunately the prime example, choline acetyltransferase, is not dealt with). The second is the production of antisera against molecules of unknown function that identify specific classes of neurones or subconstituents of synapses (this book abounds in fascinating examples). The monoclonal antisera thus generated can then be used for anatomical mapping, as reagents to study the function of the antigen or the cell that contains it, as tools for cell separation, lineage determination and a host of other uses. The editors provide a useful introductory summary of the technology of monoclonal antibody production and characterization and rome of the uses to which this new type of "magic bullet' can be put. However, I was disappointed that they did not address the very practical question of when it is appropriate to lake the plunge into monoclonal technology. For many purposes, conventional polyclonal antisera remain more effective reagents and may even be essen-
in the end convinced that this parucular approach provides the best trade-off between utility and generality. For example, it is quite easy to see, without elaborate mathematics, that phenomena ,uch as shunting inhibition can pcrforn~ logical operations such as those required for velocity detection. On the other hand. lkwmal methods of reasoning are arguably always better than verbal ones, especially in a biophysical context, and Poggio and Torte do as a result have some prediction~ tip their sleeves. Anyone who is worried by the uneas3, relationship between theory and experiment in neuroscience should find this book interesting to read, while it can also be recommended to mathematicians as a useful introduction to some areas of theoretical neurobiology. NICHOLASV SWINDALE The Physiological Laboratory. University&'Cambridge, Downing Street, Cambridge ('B2 3EG. U.K. tial, for example in producing immune lesions of specific celt types in vivo. Perhaps the inclusion of several papers using conventional antisera in this symposium on monoetonal antibodies is implicit recognition of this fact. Apart from the cost and technical problems of establishing monoclonal facilities. there are the practical problems engendered by a paucity of commercially labelled antirodent second antibodies or PAP (also protein A does not substitute nearly as well as it does for anti-rabbit second sera). More enduring problems, if the antigen is not polyvalent, include the lower potency in immunoprecipitation procedures and diminished sensitivity using indirect immunohistochemistry, either because there is only one antigen site and this may be occluded or altered by fixation procedures, or because only a proportion of the antibodies in the second serum will be directed against the specific immunoglobulin class produced by the hybridoma. Notwithstanding these minor caveats, 1 can recommend this book as an exception in the tertiary literature in that it provides a lucid and coherent exposition of an exciting new field that will be of interest to hybridoma acolytes and potential proselytes alike. LAURIEGEFFEN Professor of Human Physiology and ('onvener o] the Centrefor Neuroscience, The Hinders University of South Australia, Bedford Park, South Australia 5042.