Erickson's ambitious treatise on the idea that there are a few basic tastes challenges our assumptions about the organization of the taste system. In this regard, his arguments are thought provoking and deserve to be taken seriously. However, we offer the following comments in the spirit of joining in a lively debate about these issues.

Erickson confuses the question of whether there are basic tastes with the question of how taste stimuli are encoded by the nervous system. These are two separable issues, and their comparison is akin to comparing apples to oranges. Even if there were a “continuum” of tastes, there might still be separate groups of cells associated with each taste stimulus. So the argument that there are not (only) four or five basic tastes does not necessarily impact the labeled-line theory. Nor does it provide support for the across-fiber pattern theory since there are other mechanisms and schemes that Erickson did not consider that may also be used to encode taste.

As Erickson notes, the idea that there are only a few basic tastes goes back far in history. He argues rather convincingly that our language limits our thinking about taste and that it also guides our scientific inquiries. However, the converse could also be true. That is, it may be that our language related to taste is the result of our perception of the taste world and the brain's ability to recognize the similarity among tastes to form natural categories that logically organize our taste world. So, we may learn that sucrose, saccharin, and proline are sweet, just as we learn that collies, dachshunds, and Great Danes are dogs; they share common features. Erickson's argument that we as humans are naturally prone to organizing experiences into groups does not mean that groups don't exist.

His more cogent argument concerns the question of whether these psychophysical groups of tastes, that is, “taste qualities,” should restrict our quest for separate receptors associated with each group. Certainly, the finding of a family of bitter receptors, rather than just one (Mueller et al. Reference Mueller, Hoon, Erlenbach, Chandrashekar, Zuker and Ryba2005), argues that the taste experience may be far more nuanced than just the four or five basic taste qualities. Moreover, the recent discovery of fat receptors located on taste receptor cells (Gilbertson et al. Reference Gilbertson, Boughter, Zhang and Smith2001) may provide an impetus to broaden our definition of what a taste actually is. As Erickson correctly points out, we don't really have a formal definition of what constitutes a taste.

In his discussion of his psychophysical experiment, Erickson argues that these data falsify the idea of four basic tastes on several counts. First, his subjects could do just as well at “accounting” for the array of tastants using “non-basic” taste stimuli as they did using the “basics” (by which we presume he means prototypical exemplars of the basic taste qualities). However, his non-basics were not shown to be independent from his basics; on the contrary, his non-basics evoked taste sensations that were similar to one or more of the basic taste qualities. So the observation that subjects could do just as well using non-basics as basics is not surprising and does not falsify the basic grouping of tastes. Second, Erickson argues that since the accounting was not total there must be other tastes that can account for the missing quality. This point is related to his next argument that taste qualities were present that were not related to the basics tastes. However, examination of the other descriptors offered by the subjects only underscores the multimodal nature of taste stimuli; tastants can evoke thermal, tactile, and olfactory sensation in addition to taste.

Labeled-line versus across-fiber pattern theories

Erickson paints a dichotomy between the labeled-line theory and the across-fiber pattern theory and sets them up as a debate that is still raging unresolved in the literature. We would argue that both the dichotomy and the debate are straw men.

Consider first the differences between the labeled-line theory and the across-fiber pattern theory. These theories are by no means mutually exclusive and are therefore difficult to disentangle. Both are spatial theories in that they rely on the identities of the neural elements that produce activity. So, even if there were a labeled-line code, with different neuron types encoding different taste stimuli (basic or not), such an arrangement would obviously result in different across-fiber patterns of activity. Conversely, if the activity across all cells were the critical conduit of information, then if two patterns were different, there would necessarily be some cells that fired more for one stimulus than the other and these cells could be called labeled lines. For example, Smith et al. (Reference Smith, Van Buskirk, Travers and Bieber1983) showed that cells that responded most to sucrose in the hamster brainstem were essential for the tight association of across-neuron patterns of sweet-tasting stimuli in a multidimensional taste space. Moreover, if all cells were broadly tuned so that they participated equally in multiple discriminations, the information that they could convey would be limited: the more broadly tuned the cells, the more homogeneous the contribution to the across-fiber pattern and the more ambiguous the message. After all, as Erickson himself points out, it is the differences between across-fiber patterns that convey information and these differences must arise from cells that show differential sensitivity across tastants.

Both the labeled-line and across-fiber pattern theories use as their basic data point some measure of response consisting of the number of spikes in an arbitrary time period during which the tastant is on the tongue. This measure assumes that the activity is integrated over this time interval. The Neural Mass Index, which Erickson argues is a better indicator of similarity among tastants than the Pearson correlation also contains the implicit assumption of integration over time. In all of these conceptualizations the temporal patterns of activity and potential cooperativity among neurons are ignored, even as these measures are emerging in the literature as informative coding mechanisms in the central nervous system (Di Lorenzo & Victor Reference Di Lorenzo and Victor2003; 2007; Katz et al. Reference Katz, Simon and Nicolelis2001; Reference Katz, Nicolelis and Simon2002; Roussin et al. Reference Roussin, Victor, Chen and Di Lorenzo2008).