We focus on some similarities and differences between Lashley's and Hebb's positions regarding the reverberatory circuit, memory traces, and local versus global patterns. We also consider the implications of their views for current research, comparing some of their agreements and disagreements in the light of recent experimental findings and models.
2. Both Hebb and Lashley emphasized the autonomy of the brain, thus reacting against the stimulus-response paradigm that was dominant in that period. Orbach provides convincing arguments concerning Lashley's embracing of the notion of the reverberatory circuit in brain functional activity, which had been presented in a mechanistically detailed form in Hebb's book "The organization of behavior" in 1949. It may be difficult to judge, however, which author introduced the reverberatory circuit to neuropsychology, given that the notion was embedded in their very different perspectives. We agree with Orbach (p. 106) that due to Hebb's clearer style (Lashley was not always easy to read), he may have found a more receptive audience in the neuropsychological community with respect to the reverberatory circuit.
3. Whereas Lashley's and Hebb's views on brain functional activity were similar both with respect to the autonomy of brain activity and in their emphasis on short-term and long-term intrinsic neurodynamic patterns, the authors disagreed on the properties of memory mechanisms in terms of reverberatory activity and neural traces (assemblies). According to Lashley, memory traces are basically given by molar and redundant patterns in the cortex, based on the criteria of mass action and equipotentiality. According to Hebb, memory traces are laid down in the discrete structure of cell assemblies. He focuses on the reinforcement of a subset of individual synapses between cells of a given assembly, with strong interactions within the cell assembly and weak interactions with "external" cells. These two positions are not necessarily in conflict if we broaden the meaning of connectivity beyond the existence of neuroanatomical, connective structures.
4. We agree with Orbach's argument (pp. 8-9) that according to Lashley the diffuse nature of memory traces does not exclude a structure in terms of physiological rather than anatomical connectivity. Based on electrophysiological studies of neuronal cooperativity, a functional or effective neural connectivity, which is dynamic on different time scales, has recently been distinguished from structural or anatomical connectivity between neurons (e.g., Kruger, 1991). So one could define functional or dynamic neural assemblies, corresponding to memory traces, on several spatial and temporal scales, from local to global and from relatively transient to stable assemblies. Hebb and Lashley also differed in their emphases on different neural learning mechanisms underlying the formation of memory traces in the brain.
5. As Orbach's book illustrates with a careful selection of passages from Lashley, the latter author stressed the role of excitatory wave irradiation and selective resonance of neuronal circuits, involving at the same time a large number of spatially distributed neurons, without strong functional coupling between subsets of locally connected cells. In the light of recent neurophysiological evidence, this particular view of Lashley could be based on principles of large-scale neuronal cooperativity rather than on the functional role of individual synapses. In this perspective, neurons act as correlation detectors and wave generators. Lashley's notion of pattern interference in the brain, with the emergence of dominant resonant patterns, closely resembles the notion of cooperative modes used in the physics of complex systems and extended to brain modelling (Haken, 1973, 1996). It is difficult, however, to relate complex behavioral performances like those in Lashley's experiments to specific memory traces (see LeDoux's stance in this respect, which is discussed by Orbach on pages 67-68).
6. Hebb (1949) focuses on the reinforcement of a subset of individual synapses between cells of a given assembly, with strong interactions within the cell assembly and weak interactions with "external" cells. It seems to us that Hebb (1949) explicitly related the reverberatory circuit to the formation of memory traces in the brain, i.e. short-term memory or attentional functions, including learning and long-term memory. In contrast, the mechanism of neural plasticity has not been mechanistically described in a comparably precise manner by Lashley, nor has it been clearly related to the reverberatory activity.
7. We agree with Orbach that it is important not to forget 'Lashley's lesson' according to which synapses that are inactive during training may still show the effects of training. Indeed, as synapses between neurons are usually very weak, and pyramidal cells in the cortex are arranged in multiple parallel loops (Braitenberg & Schuz, 1991), cortical information processing simultaneously involves thousands of spatially distributed neurons and synapses. Moreover, memory traces are unlikely to be inscribed in a tabula rasa, and synapses which are involved in a given training session are functionally related to innumerable other synapses which are not directly involved. Thus, we like to think of Lashley as a pioneer of neuronal cooperativity and functional coupling among neurons.
8. Another significant viewpoint of Lashley's, rightly emphasized by Orbach, is that neurons are not inert and static; it is, for instance, precarious to compare working brain circuits with electronic circuits. Moreover, Lashley stressed the peculiarities and complexity of working brain properties and brain function as a whole. So, we may assume that if he were alive today, Lashley would criticize many simplified neural modeling studies. However, while agreeing with Lashley about the intrinsic complexity of neurodynamics, we would point out that the appropriate model-dependent simplification of neuronal interactions and properties is sometimes necessary to understand how neurons interact with each other.
9. It is not surprising that Hebb's clear notions of cell assembly (reverberatory circuit) and synaptic plasticity have paved the way for the connectionist approach to cognitive processes, inspiring many neural network models (e.g., see Amit, 1989, 1994). As Orbach argues eloquently (p. 108), while Hebb stressed the sequential activation (phase sequences) of cell assemblies, a complementary neural mechanism may be based on the parallel activation of many reverberatory circuits. In Lashley's view, a mechanism like this would be involved in the reduplication of memory traces. However, Lashley did not specify the properties of the hypothetical resonance mechanism involved in the reduplication of memory traces. A likely candidate for the formation of multiple spatially distributed assemblies via resonance may be transient coherence in gamma range oscillations. For instance, very recently Miltner et al. (1999) have shown that increased gamma-band EEG activity is involved in associative learning. Moreover, they found that another measure, gamma-band coherence, increased between regions of the brain that receive stimuli paired in an associative-learning procedure (in humans). This increase in coherence may reflect binding among neuronal activities in multiple brain regions and the formation of non-local Hebbian cell assemblies, i.e. long-term memory neural representations.
10. Lashley's anti-localist views on brain functions can be related to Anokhin's (1935) and Bernstein's (1967) stances on the relationship between central nervous system activity and behaviour. Both Russian physiologists emphasized systemic interactions in the brain. Anokhin (1935) introduced the notion of a functional system, which was elaborated in neuropsychology by Luria (1973). According to Anokhin, the functional system is an integrative formation of the organism, in which a given behavioral function is related to the concerted work of many central and peripheral components, and a given component (for instance, a cortical area) may be related to multiple functions. Like Lashley, Bernstein (1967), emphasized active physiological states and anticipatory brain activity, conceiving perception and motor coordination as operating on integrated patterns (he introduced the concept of motor synergy).
11. We will now consider the problem of integrating spatially distributed activity states in the brain. According to Lashley, holistic patterns of transient neural activity -- resonant parallel reverberatory circuits and memory traces -- are formed in the brain. Moreover, as passages selected by Orbach point out, Lashley interestingly refers to a functional superposition of activity processes on slower (tonic processes) and faster (fluctuating processes) time scales. However, some radical implications of Lashley's non-localist thesis, stressing the absolute dominance of global neural patterns, seem to be inconsistent with recent neuroanatomical and neurophysiological data. The visual cortical system, for example, exhibits the coexistence of functional segregation and integration at several processing stages (e.g. Zeki & Shipp, 1988; Tononi et al., 1992). Thus, integrative (binding) processes may occur in the presence of specialized neuronal responses, without interfering with local coding (e.g. receptive field properties).
12. On the other hand, even at the level of single cells, it is often difficult to relate a single neuron or set of neurons to a specific function (Vaadia et al., 1991). Finally, neural assemblies implementing memory traces are likely to be composed of sets of neurons in multiple cortical areas and in extra-cortical regions. For example, an interplay between sets of cortical and hippocampal neurons may be required in long-term memory storage, i.e. trace consolidation (Murre, 1996). In this light, both the radical localist (Page 2000) and non-localist views appear to be reductive (see also Murre 1992).
13. Edelman (1987) suggested that plasticity in the brain is modulated by diffuse neurotransmitters in a heterosynaptic manner, and that non- local synaptic patterns (global maps) emerge in learning as a result of widely distributed synaptic changes involving long-range re-entrant connectivity and value-dependent release of diffuse ascending neurotransmitters (also see Tononi et al., 1992). In our view, in some respects Edelman's theory implicitly relates Hebbian associative synapses (among re-entrant connections, modulated by slowly decaying concentrations of neuromodulators) and assemblies (in terms of locally connected neuronal groups), to the global patterns of activity and synapses favored by Lashley.
14. Finally, we would like to mention a possible convergence of Hebb and Lashley with respect to the cortical correlate of Lashley's "serial order in behaviour". In our view, dynamic cell assemblies of the synfire type may be involved in 'microstructuring' serial behavioral and cognitive operations, as in thought and language (also see Braitenberg & Schuz, 1991). Synfire chains (Abeles, 1991) are feedforward networks of functionally connected neurons, with diverging/converging synaptic links from one set of neurons to the next. Synchronous emission and arrival of spikes (action potentials), in other words neuronal cooperativity, is required to accurately transmit signals along the chain. We are currently modeling the self-organization of long-range cortical chains through competitive Hebbian plasticity (Murre & Raffone, in preparation). Interestingly, the amplitude and stability of neural wave irradiation processes, like those hypothesized by Lashley in trace reduplication, appear to increase with Hebbian plasticity and stochastic self-organization of synaptic strengths.
Abeles, M. (1991). Corticonics: Neural circuits of the cerebral cortex. Cambridge (UK): Cambridge University Press.
Amit, D.J. (1989) Modeling brain function. Cambridge (UK): Cambridge University Press.
Amit, D.J. (1994). 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
Anokhin, P.K. (Ed.) (1935) Problemy centra i periferii [Peripheral and central problems]. Gor'kij.
Bernstein, N.A. (1967) The coordination and regulation of movements. Oxford: Pergamon Press.
Braitenberg, V., & Schuz, A. (1991) Anatomy of the cortex. Berlin: Springer-Verlag.
Edelman, G.M. (1987) Neural Darwinism. The theory of neuronal group selection. New York: Basic.
Haken, H. (1973) Synergetics: cooperative phenomena in multi- component systems. Stuttgart: Teubner.
Haken, H. (1996) Principles of brain functioning: A synergetic approach to brain activity, behavior and cognition. Berlin: Springer-Verlag.
Hebb, D.O. (1949). The organization of behavior. New York: Wiley
Kruger, J. (Ed.) (1991) Neuronal cooperativity. Berlin: Springer.
Luria, A.R. (1973) The working brain: An introduction to neuropsychology. New York: Basic Books.
Miltner, W.H.R., Braun, C., Arnold, M., Witte, H., & Taub, E. (1999) Coherence of gamma-band EEG activity as a basis for associative learning. Nature, 397, 434-436.
Murre, J.M.J. (1992) Precis of: Learning and Categorization in Modular Neural Networks. PSYCOLOQUY 3(68) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.3/ psyc.99.3.68.categorization.1.murre http://www.cogsci.soton.ac.uk/psyc-bin/newpsy?3.68
Murre, J.M.J. (1996) TraceLink: A model of amnesia and consolidation of memory. Hippocampus, 6, 675-684.
Murre, J.M.J, & Raffone, A. (in preparation) Modeling self- organization of long-range multisynaptic circuits in the cerebral cortex.
Orbach, J. (1998) (Ed.) The Neuropsychological Theories of Lashley and Hebb. University Press of America
Orbach, J. (1999) Precis of: The Neuropsychological Theories of Lashley and Hebb. PSYCOLOQUY 10(029). ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.029.lashley-hebb.1.orbach http://www.cogsci.soton.ac.uk/psyc-bin/newpsy?10.029
Page M. (2000) Connectionist Modelling in Psychology: A Localist Manifesto Behavioral and Brain Sciences 23 (4) (in press) http://www.cogsci.soton.ac.uk/bbs/Archive/bbs.page.html
Tononi, G., Sporns, O., & Edelman, G.M. (1992) Reentry and the problem of integration of multiple cortical areas: Simulation of dynamic integration in the visual system. Cerebral Cortex, 2, 310-335.
Vaadia, E., Ahissar, E., Bergman, & H., Lavner, Y. (1991) Correlated activity of neurons: a neural code for higher brain functions? In: Neuronal cooperativity (J. Kruger, ed.), pp. 249-279. Berlin: Springer.
Zeki, S., & Shipp, S. (1988) The functional logic of cortical connections. Nature, 335, 311-317.