Daniel J Amit (2000) Hebb Versus Lashley: the Practitioner's Case. Psycoloquy: 11(008) Lashley Hebb (10)

Volume: 11 (next, prev) Issue: 008 (next, prev) Article: 10 (next prev first) Alternate versions: ASCII Summary
Topic:
Article:
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 11(008): Hebb Versus Lashley: the Practitioner's Case

HEBB VERSUS LASHLEY: THE PRACTITIONER'S CASE
Book Review of Orbach on Lashley-Hebb

Daniel J Amit
Dept of Physics,
Universita' di Roma, La Sapienza
http://racah.phys.huji.ac.il/~damita/

amit@jupiter.roma1.infn.it

Abstract

Lashley made important contributions but was less prescient than Hebb, especially concerning the importance of synaptic learning and localization. On balance, it is Hebb's view that has prevailed.

Keywords

cell assembly, central autonomous process, engram, equipotentiality, Hebb, Hebbian learning, Lashley, localization, memory trace, nativism, reverberatory circuit, Vanuxem Lectures
1. This book (Orbach 1998) is a treasure. It is highly recommended for anyone interested in neuroscience or in intellectual history. It is a book with a great passion for truth and for justice, both historical and scientific. Science would greatly benefit from more books of this type. Besides, it is an invaluable compendium of Lashley's writing, in particular the unpublished Vanuxem lectures. I feel much rewarded for the experience of reviewing it. Yet I cannot accept its central thesis.

2. Orbach (1998, 1999) suggests that theoretical neurophysiology and psychology of the last 50 years have been guilty of marching under the banner of D. O. Hebb, whereas the major ideas were in reality developed and expounded earlier by K. S. Lashley, Hebb's teacher. Orbach attributes the wide promulgation of this injustice to defects in Lashley's prose, but this causal attribution raises itself serious doubts about the strength of Orbach's case. Reading Lashley for the first time, I find his style and scope intoxicating. Indeed, his arguments are so powerfully phrased that they tend to convince even where they completely miss the point, as in dealing with epistemology.

3. In a nutshell (but not intended as a substitute for reading at first hand this masterly treatise and the unusual running commentary of Orbach interspersed among the papers of Lashley reproduced in the volume), the central issue revolves around priority in the identification of the explosive significance of Lorente de No's neural feedback circuit. Such circuits, it was realized, could sustain the effect of a current stimulus for later reflective processing by the brain. This possibility contrasted sharply with the prevailing "primitive" behaviorism of Watson, which focused exclusively on current stimulus-response sequences. The book leaves no doubt that Lashley was the first to note these implications with magnificent arguments that should have laid to rest the behavioral psychology of the day, dominated as it was by Pavlovian concepts. Yet they did not.

4. In reality, reverberating circuits have not had a major impact on psychological thinking, nor on computational neurophysiology. I have argued (Amit 1995) that the Hebbian paradigm has been perceived mainly as a message about synaptic engineering, to be used in computing "reasoning" in feed-forward neural mechanisms. Such networks are still of the stimulus-response type. If one seeks a simple explanation of the prevalence of Hebb in the collective consciousness of the community, it is to be found precisely here, in his promotion of learning.

5. This idea was anathema to Lashley, and it appears that Orbach is unwilling to assign a penalty for this blindness. So much of Lashley's Vanuxem Lectures is dedicated to making the idea of learning -- the formation by experience of structured, long-lasting cortical structures -- appear contrary to experiment. I would go as far as to speculate that the reason Lashley did not mention the contents of Hebb's (1949) book, which he had seen, is that he was unwilling to legitimize the approach based on plasticity, which Lashley must have perceived to be the central message of the book. Whether or not Lashley's position on (opposition to) synaptic plasticity will ultimately prove correct is still an open question. However, a proper history of science would note that the Hebbian formation of neural assemblies through such plasticity, in response to stimuli, has fuelled a major "research program," in Lakatos's terminology; whereas Lashley's perspective appears in retrospect, to a practitioner, rather barren.

6. If one takes a broader view of the issues involved, one must note that Lorente de No's circuit has never been more than a metaphor. Neither then nor now has a cortical circuit of this simplicity actually been shown to produce "reverberations." Lorente de No's observations were mostly anatomical (structural), not physiological (functional). It is in fact rather unlikely that circuits as small as those he observed could actually reverberate. The important insights, beyond the metaphor, are that (1) feedback is observed in neural circuitry; (2) feedback is essential for reverberation; and (3) reverberations are necessary to go beyond behaviorism. These facts were clearly noted by Lashley.

7. Where Lashley's position becomes a burden for the development of a systematic understanding of the implications of the reverberatory idea is (a) his resistance to synaptic learning and (b) his resistance to localization. Major trends in computational neuroscience, from the Perceptron via PDP to reverberating Hopfield-type assemblies (see Amit 1995) have been fuelled by the idea of learning from experience encoded in localized synaptic structures. Learning was essential for the psychological dimension, as well as for artificial intelligence. Learning is rightly seen as a way for a neural system to change its repertoires. It has turned out to be a productive paradigm, with no serious competition. In fact, when learning was shown to be limited in the context of the single-layered Perceptron, the subject went into a deep (but temporary) crisis (Minsky & Papert 1988).

8. The foregoing comments represent a possible resolution of the main issue that the book leaves in a rather unsatisfactory state. They suggest why and how Hebb, rather than Lashley, has provided the building blocks for modern computational neuroscience. Whether or not Hebb was right remains a scientific rather than a psychological question. The final verdict is not yet in. But it can already be said with reasonable certainty that if Hebb's paradigm is rejected, Lashley's picture will not provide the alternative. Despite its great elan and imagination, it lacks constructive elements.

9. To conclude let me offer a short and partial balance sheet of some less central, yet quite interesting, issues.

10. Pro Lashley:

    a. Any nerve cell may participate in many patterns of reverberating
    activity, and the behavior is statistical (p. 287). This
    encompasses the idea that latencies for different items can reside
    in different combinations of cells in the same area. They can also
    act together (p. 324-325).

    b. Many cells must participate in every reverberation: the
    influence of one cell is too slight to detect (ibid).

    c. Priming of cells by ongoing activity is related to the timing of
    responses to stimuli (p. 289). This fundamental idea has not yet
    been fully assimilated, as indicated by Thorpe et al. (1996).

    d. The idea of the priming of cells by ongoing activity as a
    mechanism for selective attention (p. 290) is beautiful and
    most-modern.

    e. It is much to the credit of Lashley to have had the intuition
    that the heavily formalized proposals of McCulloch & Pitts (1943)
    had little bearing on either mental computation or the solution of
    the problem of "stimulus equivalence," which we now call invariance
    (p. 312).

11. Contra Lashley:

    a. Lashley's priming is on single cells -- Hebb sees it on the
    assembly level.

    b. Hebb's attentional idea, where the activity in one assembly is
    at the origin of the type of response to a stimulus of another, is
    very modern and different from Lashley's.

    c. Lashley's attitude to inhibition, as interfering excitation, is
    unreasonable (p. 314). Alhough at some point (p. 286) explicit
    inhibition is invoked, the metaphor for inhibition is that of two
    successive spikes to the same cell causing inhibition. It appears
    as part of a general importation of a linear metaphor from wave
    theory (pp. 28, 290).

    d. The idea of reduplicated memory traces (p. 314) is rather
    peculiar. Lashley himself says: "I fear there is very little to be
    said in favor of this theory." Yet he does not relinquish it.

REFERENCES

Amit, D.J. (1995). The Hebbian paradigm reintegrated: Local reverberations as internal representations. Behavioral and Brain Sciences 18(4):617-626. http://www.cogsci.soton.ac.uk/bbs/Archive/bbs.amit.html

Hebb, D.O. (1949) The Organisation of Behaviour. New York: Wiley.

W. S. McCulloch & W. Pitts (1943) A logical calculus of the ideas immanent in nervous activity. Bulletin of Mathematical Biophysics, 5: 115-133.

Minsky, M.L & Papert, S. (1988) Perceptrons : An Introduction to Computational Geometry. MIT Press.

Orbach, Jack (1998) The Neuropsychological Theories of Lashley and Hebb. MD: University Press of America.

Orbach, J. (1999) Precis of: The Neuropsychological Theories of Lashley and Hebb. PSYCOLOQUY 10(23). ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.029.lashley-hebb.1.orbach http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.029

Thorpe, S., Fize D., & Marlot C. (1996) Speed of processing in the human visual system. Nature 381(6582): 520-522.


Volume: 11 (next, prev) Issue: 008 (next, prev) Article: 10 (next prev first) Alternate versions: ASCII Summary
Topic:
Article: