Ramachandran’s “The Tell-Tale Brain”

I am commencing Ramachandran’s newest book after having read “A Brief Tour of Human Consciousness” and “Phantoms in the Brain”. Even his preface is entertaining. He points out that his clinical methodology would have been available to Hippocrates. This relates, I imagine, to the reason that I find Ramachandran’s style of theory satisfying: He explains things in terms of causal chains involving, simpler entities. ‘Simpler’ is not an entirely objective notion. Freud’s explanations are often in terms of things that he might have thought simpler, but are entirely unfamiliar and inscrutable to me. Different people ‘bottom out’ in different sorts of things. Ramachandran and I both bottom out in inanimate things.


Ramachandran considers metaphors seriously, suggesting several mysteries. Here are my thoughts on metaphors. It seems perfectly clear why metaphors are useful; they help us see patterns when distinct things in the world are like each other in some way. This broaches two serious mysteries:
Why things are like each other?
I see these possibilities: I vote for the first.
How our brain exploits these metaphors on a neural level
Grossly we store mappings from one domain to another and carry learning in one domain to the other, but how do we do this?
Metaphors may be the brain’s pattern tool—no more; no less.

There is a problem with the simple cross-wiring hypothesis for synesthesia. Figure 3.4 has intermixed congruent 5’s and 2’s. The report suggests that the digit field is processed concurrently—i.e. in parallel. The ability to recognize off axis characters requires replicated hardware as any OCR engineer will tell you. I assume here that there is no panning hardware other than eyeball movement. The cross wiring would have to systematic so as to apply to the multiple processing elements. I believe that we have parallel edge detectors, constructed by non duplicated genes. I don’t believe that we have parallel ‘7’ detectors. Happenstance mis-wiring is unlikely to make 7’s red in the whole field, as reported.

Metaphors are shared culturally and are often adaptive. Synesthesia, as reported by Ramachandran, seems to consist of non-adaptive sensory associations, or functions (sets of associations) between two specific domains, depending on the individual. Ramachandran does not suggest that it is often adaptive.

Metaphors use words in inexact ways. Concerning the design of complex hardware or software systems:

My Number Line

I have an extremely conventional number line. I think that I remember some problems with the teens, and fitting them into the numbers less than 100. I recall from an early age the pattern of two digit numbers. I think that numbers have no spatial individuality to them. When asked to compare two 2 digit numbers, I certainly don’t locate them on a number line; I consider their respective Arabic numerals. It is hard for me to imagine having learned numbers via Roman numerals.

I do not consider the early selection of words (assigning names to things) to be a problem. I grant the bouba-kiki phenomenon; I experienced it myself and I suppose it needs an explanation. It seems unnecessary to explain ‘the first words’. Nature finds symmetry breaking easy, once the need arises. Systems in unstable equilibria just don’t stay that way long.

The bouba-kiki phenomenon seems plausibly cultural to me. I wonder if it has been checked across cultures.

Mirror Neurons

Ramachandran provides much information new to me on mirror neurons. He convinces me that they will play a central rôle in adequate future theories of the evolution of the mind—he asks the right questions. I am not impressed with his speculations however but I think his observations and speculations are highly productive.

The mirror neurons are the tip of some iceberg. I think more of the iceberg must be seen to settle some of Ramachandran’s speculations. That an image of a massaged hand relieves pain in a phantom limb suggests that mirror neurons supplement the ‘normal’ signals from the body to the brain. Is this utility great enough to partly explain some of their origin? Ordinary Hebbian learning might explain such cross signals. These ideas must explain how recently mirror neurons appear in our lineage.


The information on Wernicke’s and Broca’s areas is also new to me and highly relevant. The last 8 pages of chapter 6, on language, are perhaps the best of the book. He addresses the tangled relations between speech and thought, and the brain parts where they seen to happen.


The paper by Wolf and Gray as described (page 205) by Ramachandran sounds very significant to me. The title is “Visual feature integration and the temporal correlation hypothesis”. I cannot find it online. This is perhaps a followup article by Gray. I smell a link to Kanerva’s ideas, but I can’t see it yet.

Page 221:

This may be true for power supply reasons, but not for logical reasons. I would imagine that ‘attention’ would need as much of the ‘general purpose’ brain as it could muster. The limitations are power and prior commitments. These are my intuitions as a computer person. While the brain is not a computer we need to be clear why we come to different conclusions than apply to the only partially brain like model that we do understand.

There are two ‘laws of aesthetics’ that seem to conflict: (6) Abhorrence of coincidence, (8) Symmetry. I think there is an easy way out: The objectionable coincidence is in the raw 2D image and the desired symmetry is in the 3D shape deduced from the observations. A snowflake is perceived as a symmetric 3D which happens to lie in a plane, itself a form of symmetry.

Ramachandran introduces Jason at the beginning of chapter 9. Jason seems to be visually disconnected from the world, or at least consciously unaware of it. Jason speaks coherently on the phone with his father. I want to know what happens if his father asks whether Dr. X is in the room with him. These are the sorts of questions that Ramachandran is normally likely to ask himself and that is what makes his reports so useful.

I like a section on consciousness starting on page 247 that begins:

I think that this is a good ontological stance. I would add that our view of ourselves is merely a new sense which, like other senses, has evolved because it is adaptive. The new sense had a head start (pardon the pun) for already having ready neural access to the subject matter. It explains why dualism seems attractive. It even provides a vantage point from which to describe, and perhaps reject solipsism. See this.

I am disturbed by phrases such as “the inhibitory circuits which ordinarily keep mirror-neurons activity in check” (page 272). I do not dispute the problem of distinguishing self from other signals and that nature has somehow solved it after a fashion. As a software engineer and sometimes hardware kibitzer I know that such mechanisms are kludges that solve the simple cases and screw up on the more complex cases, often in ways that cannot be further kludged around. I do not have a principled description of such screw-ups, merely an aversion to such designs. This may bear on the pathologies that Ramachandran describes. If this is the way that nature avoids self-other confusion then software engineering insights may have something to add. More later.

I am distressed that Ramachandran does not speak of memory. There is, however, a fascinating glossary entry under ‘semantic memory’ where he reports that common names for things seem to be the root of other associations with those things. If we forget the name of something, we are most likely to forget those things that go with it. This is very significant! This is presumably why people do arithmetic in their first language.
I see a significant problem: We quickly locate short slanted line segments among a field of vertical segments. We even seem to locate them with our peripheral vision. I presume this starts in the edge-detector area. I think that this strongly indicates parallel processing, unless there is some sort of fairly fast panning mechanism other than movement of the eye-ball. Indeed such parallelism sounds adaptive and achievable via the genes. I cannot imagine such a mechanism being learned, as seems necessary to explain the synesthete’s ability to find learned 2’s among the field of 5’s according to their ‘color’.

See a vague outline of solution to puzzle.


Starting on page 250 he mentions 6 aspects of self: