A Hebbian mechanism is a relatively simple concept that is reminiscent of behaviourism in a number of ways. The theory, proposed by Donald Hebb in 1949, states that when an axon of one cell repeatedly stimulates another cell, in time the efficacy with which the former stimulates the latter is increased via some essential metabolic change on the part of one or both of the cells involved. This idea has since been supported by a wealth of empirical evidence. To an extent the concept of Hebbian mechanisms reflects on a biological plane the more theoretical concept of classical conditioning; the two cells become associated with one another in a similar way to a conditioned stimulus becoming associated with an unconditioned response.
The first real demonstration of actual biological change being brought about by training came from Bennet et al. (1964, as cited by Rosensweig, M.R. et al.). In Bennet’s investigation rats were kept in three conditions, these being standard (a normal sized cage for three rats) impoverished (the same sized cage but with only a single rat) and enriched (where up to ten or twelve rats were kept in a large cage containing a number of stimulating objects which were altered daily). The enriched condition was considered to provide significantly higher opportunities for informal learning than the standard condition. After a period in these different conditions the rat’s brains were dissected and analysed, at which point it was found that the rats in the enriched condition had significantly higher activity of certain enzymes connected with synaptic transmission than the rats in the impoverished condition, presumably due to their greater learning experiences. The weight of the enriched rat’s brains had also increased, and this was attributable to a noticeably denser cerebral cortex. The rat’s brains seemed to have fundamentally altered in structure due to their experiences. However, since the rats had been dissected there was no 100% accurate way of knowing exactly which features of their biological change had affected what aspects of their mental abilities or what their newly bolstered cells were capable of.
The specific cell alteration supposedly brought about by Hebbian mechanisms is known as long-term potentiation. It is theorized that long-term potentiation (or LTP) may form the core of all learning and memory within the brain. Researchers working with rabbits (Bliss and Lomo, 1973) found that LTP could be observed in action: when one neuron is stimulated and the electrical output of a connected neuron is recorded, low frequency stimulation results in little change in activity over time. However, if a tetanus is administered to the input neuron then future output from the connected neuron is increased in magnitude. The large number of action potentials provided by the tetanus in just a few seconds ‘prime’ the connected neuron to work more effectively in future. This is how LTP works: repetitive stimulation of one neuron by another results in an enhancement of the synapse between these neurons. These findings have since been replicated many times with cells in culture, adding to the validity of the concept.
LTP can also effectively explain behavioural or associative learning. When a tetanus is applied to a weak neuronal input then no LTP takes place. But when a tetanus is applied to a weak input and a strong input simultaneously, LTP takes place for both neuronal inputs. As a result, a strong input, when activated simultaneously with a weak input, can actually increase the efficacy of the synapse between the weak input and the target neuron. This may be the biological foundation of behavioural learning, and it finds its roots in Hebbian mechanisms. In this example the tetanus provides the repetitive stimulation which results in increased synaptic efficacy between pre and post synaptic neurons.
In order to fully investigate LTP as the basis for behavioural or perhaps even all learning there are four criteria that need to be fulfilled, as suggested by Martin and Morris, 2002. First, any behavioural learning exhibited by an animal should result in some detectable biological change in it’s nervous system. If behavioural learning can take place without this, LTP cannot be it’s cause. In order to test this factor Murphy and Glanzmann (1999) induced LTP in Aplysia, a marine invertebrate ideal for study because of it’s simple nervous system. First the strength of a specific sensory to motor synapse was tested and recorded. Classical conditioning was utilized to teach the Aplysia, which withdraws it’s gill only when it’s tail is stimulated, to withdraw its gill when it’s mantle was stimulated. After the conditioning had taken place the strength of the recorded synapse had undergone definite learning related changes. It was also found that these changes were dependant on NMDA, the particular glutamate receptor upon which LTP relies. This is clear evidence in support of Hebbian mechanisms as the source of behavioural learning, but it must be borne in mind that the Aplysia is a very basic life form and holds nothing even approaching the level of sophistication of a human brain. That being said, demonstrating that behavioural learning is biologically possible via LTP in any animal is a valid discovery and the composition of nerves is essentially the same across species, though their structure may differ.
Martin and Morris’ (2002) second criteria, mimicry (that if the neurons could have their make-up altered surgically, the animal to whom they belonged would display learning and memories of events which did not occur) remains the subject of science fiction for the moment, but the third, anterograde alteration, is perceivable. This entails blocking the process of LTP, which if LTP is to be considered the cause for learning and memory should stop both these phenomena from occurring. Morris (1989) found this to be exactly the case in a study that examined spatial learning in rats. The rats were placed in a large tank of water, somewhere in which there was a platform just beneath the surface that the rats could rest on. The rat’s natural response to entering the water was to search for dry land, and to begin with their searches were based on trial and error. However after a certain amount of trials the rats would learn where the platform was and would swim to it almost immediately, or at least spend more time looking for it in the correct section of the pool than any other section. When Morris (1989) blocked the rats NMDA receptors in the hippocampus using a chemical called AP-5, no matter how many trials the rats entered the pool for they never learned to narrow their search to the area the platform was in. Their spatial learning had been inhibited by their neurons inability for LTP, which demonstrates in a very basic way how simple trial and error learning is dependent on Hebbian mechanisms for it’s success. Once again however, these experiments were performed on non-human animals which reduces their validity. The study also tests only inhibition of spatial learning and behavioural learning in it’s most basic form. In order for these findings to be conceivably applicable to humans, more sophisticated forms of behavioural learning had to be proved to be repressed by lack of LTP.
Attwell, Rahman, & Yeo (2001) demonstrated a similar pattern in rabbits with reference to classical conditioning, one of the cornerstones of behavioural learning. It was found that both acquisition and extinction of a conditioned response to an auditory tone (in this case blinking of the eye, with the unconditioned stimulus being an airpuff to the eye, the unconditioned response being the blinking reflex, the conditioned stimulus being the auditory tone and the conditioned response being a pre-emptive eyeblink) could be effectively halted by deactivating the cerebellum. At this point there seemed to be no specific link between the inhibition of classical conditioning and lack of LTP as the entire cerebellum had been deactivated: there was no way to be sure which specific section was responsible for learning. It has since been suggested, however, that during classical conditioning the efficacy of the synapses between Purkinje cells and parallel fibres in the cerebellum is increased, i.e. that Hebbian mechanisms are in effect, and that these mechanisms are also likely to form the cerebellar basis of other forms of motor learning.
Evidence for criteria four in Martin and Morris’ (2002) list (retrograde alteration) comes from a study by Attwell, Rahman, Ivarsson & Yeo (1999) which showed that alterations of synaptic weights after a conditioned response had been learned (the same eyeblink reflex) could prevent the expression of a conditioned response immediately (i.e. there was no period of extinction).
Despite the certainty with which many psychologists perceive the existence of Hebbian synapses and the strength of the evidence that supports Hebbian mechanisms as the source of behavioural learning, there are some who argue against it. LTP is a non-specific biological alteration in the efficacy of a synapse; some psychologists theorize that the change that brings about learning and memory is due to alterations in the fundamental structure of neurons, in opposition to the popular connectionist theory. Possibly the most compelling evidence for the role of Hebbian mechanisms in behavioural learning, particularly classical conditioning, is the reliability of connectionist principles with regards to learning and memory to which LTP is critical. Connectionist ideas suggest that the engram (or the location of memory) is not a specific location in the brain but in fact found in the web of connections formed across and between brain sections. Evidence against this is found in Thompson (1990) however, who’s investigations into classical conditioning in mammals showed that although during conditioning pyramidal neurons in the hippocampus increase in excitability in a manner that closely resembles LTP, the hippocampus is not integral to the classical conditioning process whereas the cerebellum is. In effect, the LTP taking place in the hippocampus may not only not be the cause for behavioural learning; it may not be related at all (if the engram is in fact exclusive to the cerebellum). LTP in the cerebellum still takes place during the conditioning process however, so although not all LTP that takes place during behavioural learning is key to the process, at least some seems to be.
In addition to this there are questions regarding the practical effects of LTP in a human brain. From a reductionist perspective one might argue that LTP, if it is responsible for behavioural learning, could be the building block for more complex forms of learning and memory. Harvey, Solovyova & Irving (2006) recently found that the introduction of the hormone leptin to short-term potentiated synapses resulted in long-term potentiation, raising a series of comparisons to short-term and long-term memory. Questions like this go far beyond the normal perception of Hebbian synapses as simply very basic building blocks.
In summary, overwhelming evidence indicates that Hebbian mechanisms are the basis of behavioural learning, but technological and ethical limitations mean that at least in relation to humans it cannot be accepted as irrefutable fact.
Attwell, P.J.E., Rahman, S., Ivarsson, M. & Yeo, C.H. (1999) Cerebellar cortical AMPA-kainate receptor blockade prevents performance of classically conditioned nictitating membrane responses, Journal of Neuroscience, 19, 1-6
Attwell, P.J.E., Rahman, S. & Yeo, C.H. (2001) Acqusition of eyeblink conditioning is critically dependant on normal function in cerebellar cortical lobule HVI, Journal of Neuroscience, 21 (15), 5715-5722
Bliss, T.V.P. & Lomo, T. (1973) Long lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path, Journal of Physiology, 232, 331-356
Harvey, J., Solovyova, N & Irving, A. (2006) Leptin and it’s role in hippocampal synaptic plasticity, Progress in Lipid Research, 45 (5), 369-378
Martin, S.J. & Morris, R.G.M. (2002) New life in an old idea: The synaptic plasticity and memory hypothesis revisited, Hippocampus, 12, 609-636
Morris, R.G.M. (1989) Synaptic plasticity and learning: Selective impairment of learning in rats and blockade of long-term potentiation in vivo by the N-Methyl-D-Aspartate receptor antagonist AP-5, Journal of Neuroscience, 9 (9), 3040-3057
Murphy, G.G. & Glanzman, D.L. (1999) Cellular analog of differential classical conditioning in aplysia: Disruption by the NMDA receptor antagonist DL-2-Amino-5-Phosphonovalerate, Journal of Neuroscience, 19 (23), 10595-10602
Rosensweig, M.R. et al. (2004) Biological psychology: An introduction to cognitive and behavioural neuroscience (4th edition), Sinauer Associates, Inc, 18, 556-562
Thompson, R.F. (1990) Neural mechanisms of classical conditioning in mammals, Philosophical Transactions; Biological Sciences, 329, 161-170