The following diagnostic criteria were developed by Cytowic (1):
1. Synesthesia is involuntary but must be elicited
2. Synesthesia is projected.
3. Synesthesia perceptions are durable, discrete, and generic
4. Synesthesia is memorable
5. Synesthesia is emotional and noëtic.

        This diagnosis was established because, like certain disorders such as temporal lobe epilepsy (TLE), synesthesia’s symptoms are so distinctive that they can be readily identified through description alone.  This is not the only quality that the two conditions have in common.  Indeed, temporal lobe seizures have themselves been known to trigger synesthetic experiences about 4% of the time.  Often they produce an elementary experience for example, a bitter taste, but when the seizure spreads from the hippocampus to the cortex of the temporal lobe, it tends to result in more specific and elaborated perceptions such as: "rusty iron," "oysters," or "an artichoke."
        Synesthesia is known to be heritable, though it is not clear whether it is autosomal or X-linked dominant.  It is most common in women and in non-right-handers.  Synesthetes frequently also have excellent, and sometimes eidetic memory, which they often attribute to the unforgettable vividness of their synesthetic sensations.  Overall, their cognitive skills have been described as “uneven.”  Many have subtle mathematical deficiencies, allochiria, and a poor sense of direction (5).
        Little is understood about synesthesia’s etiology.  Several studies have been performed, for instance, in one case study, a patient had audio-visual synesthesia caused by a tumor in his left temporal lobe; the synesthesia disappeared upon removal of the tumor (6).  Functional imaging techniques have also been applied, and have revealed that in MW, for example, cortical metabolism dropped 18% (below its already are low and inhomogeneous levels) on average in the left hemisphere during synesthesia.  Regional blood flow also decreased in response to amyl nitrate administration, which intensifies the synesthetic experience.  Based on these, Cytowic has concluded that it “depends only on the left-brain hemisphere and is accompanied by large metabolic shifts away from the neocortex that result in relatively enhanced limbic expression.” (5)
        Several models for the condition have been postulated.  The Cross-Modal Transfer (CMT) hypothesis suggests simply that objects can be represented, especially in infants, in an abstract form, and one that is not necessarily limited by modality.  This has been supported by findings that babies can distinguish two objects by their appearance, even if they have previously only touched them (Rose, Gottfried, and Bridger (1978); Meltzoff and Borton (1979), reviewed in (7)).  Another study (Lewkowicz and Turkewitz (1980), also in (7)) presented 1 month olds with a patch of white light followed by a burst of white noise, and found that heart rate changed the least when the intensities of the two were closest to each other (as rated by adults).  These findings contradict Piaget’s claim that sensory systems are independent at birth and gradually integrate with one another over development.
        Maurer (8) reports that widespread cortical responses to visual stimuli have been observed in infants 2 month old and younger, which suggests that “primary sensory cortex is not so specialized in the young infant as in the adult” (p. 111).  Consequently, she adheres to the controversial Neonatal Synaesthesia hypothesis, in which all humans are born with a form of synesthesia.  That is, at birth our senses have not yet developed to the point where we may experience differentiated sensation; sounds trigger not only auditory experiences, but visual and tactile ones as well.  This is an extension of the CMT hypothesis, and suggests that the cross-modality results in a sensory confusion for the infant.  An anatomical basis for this has apparently been found in other species, namely, neonatal kittens and hamsters respectively have transient connections between visual, auditory, somatosensory, and motor cortex and between the retina and the main somatosensory and auditory nuclei of the thalamus.  If the NS model does indeed hold true in these species, the human data mentioned above suggest that it may also apply to us.