Rumors are abroad that psychophysics is dead but this writer believes they have no foundation. The reasons for this belief can be laid out at two levels, a broader conceptual and a more immediate methodological.
To begin with the second, purely methodological one. Experimenters analyzing the visual system most often tackle the apparatus first. A celebrated example of this approach is Hartline (
1928), as genuine a scientist as we had in the field, shining light into the immobile eye of the horseshoe crab
Limulus and recording electric activity in a nerve fiber originating in the eye. All that was demanded of
Limulus was that they sit still, something that comes naturally to this creature. The destination of the fibers from the eye and the possible effect on
Limulus behavior did not arise until decades after Hartline had returned from Stockholm when it was reported that a male
Limulus could use vision to distinguish a female from a rock (Barlow, Ireland, & Kass,
1982). What was obvious and needs no explicit articulation is that Hartline conducted experiments on the physiological operation of structures in the visual stream, a different line of endeavor than essays into the understanding of how we see. Concentrating on transactions taking place in structures known (or suspected) to be involved in the process of vision has been an enormously successful undertaking and, with the development of new and amazingly powerful tools, retains the continued promise of new discoveries. However, even with immense effort and success, the circle hasn't completely closed.
Out of many, here is a particular example. One of the most prominent examples of selective visual processing at an early stage of neural elaboration is that oriented visual contours are a “trigger feature” of neural units in the primate primary visual cortex. That concept and details of the involved apparatus continue to occupy generations of visual neurophysiologists. However, in
1861 Mach had already measured subjects' precision in matching the orientation of contours, in
1892 Jastrow found the threshold to be of the order of 1° (as compared to the width of neurons' tuning curves of tens of degrees), and in
1913, Blachowsky demonstrated the primacy of contours in the visual field. Thus, the revelation of oriented receptive fields, justifiably regarded as a significant milestone, had not been necessary to our understanding of how we see. That it is not even sufficient is shown by excellent thresholds in the discrimination of the orientation of contours that aren't shown at all, but merely cued in.
The enterprise of reductionist visual neurophysiology is too successful, its results too fascinating, and its attraction too well matched to the contemporary view of science, to stem or even to slow. If it is thought, however, that it can proceed unimpeded toward an understanding of how we see without reference to the actual experience and performance of the organism, the progress of science does not support such a stance. If, from a half a century of reductionist effort in the visual stream of primates, we still cannot deduce performance characteristics, such as the oblique effect or hyperacuity of contour orientation, and have to rely on threshold responses of observers to outline processing operations, it is a mistaken notion to even imagine psychophysics being dead.
Talking about threshold responses of a human observer, in which a subject in what Brindley (
1960) called Class A experiments is constrained to merely a binary state, there is sometimes a hesitation to accept this as a scientific datum as valid as, say, a trace on an oscilloscope, the strength of a voxel in the readout of an MRI scan, or the decision of the edge location in the light distribution in a microscope image. A thoughtful examination of the source and certainty of knowledge gained from these modes of experimentation should remove any hesitation. If the yes/no response of an observer in a carefully controlled psychophysical experiment fails to convince of its rigor, what about the selection of which neuron to record from or include in an ensemble, the setting of the height of discriminating windows, or the choice of the color palette in an MRI display?
There is, however, an entirely different level on which to conduct this colloquy. At that level, psychophysics is not only alive and well and an important component in current vision research, it is immortal. Neuroscience's recent triumph in elucidating the olfactory system is proceeding from success to success, but in the end, the performance and richly furnished odor world of a beagle is still beyond it, and the same can be said about the vibrissal sense and the space and texture of the surround to a rat. The observer's experience is still the touchstone of vision research, it informs the end result. As yet, it has always transcended laboratory findings on components of the stream, and whether the two will ever meet has exercised philosophers' minds since they started arguing.
The giants in our field touched on the problem but mostly fail to serve as final arbiters. Fechner, of course, is the originator and by aiming at a functional relationship between them, gives equal room to the material and mental worlds. Helmholtz, the most consequent thinker, after pursuing the reductionist analysis as far as it could be taken at his time, becomes vague and has to resort to inference, albeit unconscious. Hering, on the other hand, is explicit and articulate: writing in 1870, he likens the neurophysiologist to someone laboriously examining the stage mechanisms and the movement of the actors from behind the scenery; the sense of the play, immediately obvious to the audience, eludes him (Hering,
1905). A founder of gestalt psychology, Max Wertheimer (
1923), remains wholly one-sided. Defining gestalten, he posits that they “obey structural rules and not run counter to them.” But with nearly a century of trying, such rules have yet to be found in the realm of geometry or the world of physics. Modern neurophilosophers generally urge the contrary approach: look to neuroscience to close whatever gap there remains between its findings and the ultimate personal accounts of observers' experiences (Churchland,
1986).
There is consensus. Dating from before and remaining alive throughout the successful reign of what Hering (
1905) called the “physiologist as physicist” there has been the need for, in his words, “the physiologist's indispensable assistance of psychology” (p. 6). From the very beginning, the axiom of a straight line, vision researchers have brought to the endeavor items from the toolbox originating in their own experience; their meticulous and now more and more exhaustive account of the activity behind the scenery requires supplemental information from viewing the play.
For the vision scientist, psychophysics is not only alive and continuing to serve as an active and important part of the armamentarium, but its future as an inescapable and permanent ingredient of the enterprise remains unchallenged.
Commercial relationships: none.
Corresponding author: Gerald Westheimer.
Email: gwestheimer@berkeley.edu.
Address: Division of Neurobiology, University of California, Berkeley, CA, USA.