The tendency for low latitude (tropical) populations to have larger cranial volumes than high latitude populations is consistent with a thermal stress explanation: high latitude populations are forced to divert energy away from growth into thermoregulation. Thermoregulation is known to be costly for mammals (Mount 1979). Within this overall pattern, the sudden upswing in cranial volume at around 400 ka is consistent with a rapid release from the environmental constraint. The most likely explanation is the control of fire. Fire would have introduced two crucial effects. One is to raise ambient temperatures in the immediate vicinity of living sites (and especially so in caves and other shelters), thus allowing a significant saving in the energy required for thermoregulation, especially overnight when ambient temperatures are at their lowest even in the tropics. The second is cooking, which has a significant effect on nutrient (and especially energy) extraction rates, particularly for red meat and carbohydrate-dense plant foods like roots and tubers (Wrangham et al. 1999; Wrangham 2010; Carmody & Wrangham 2009). This is not to suggest that earlier hominins never cooked, but rather to suggest that 400 ka may demarcate a phase shift in the quantity of cooking: prior to ~ 400 ka, cooking may have been opportunistic and rather casual, but after 400 ka it became habitual.
The archaeological evidence clearly suggests that, from about 400 ka onwards, hearths become a regular feature of hominin occupation sites, but that they were very rare prior to that date and their use is likely to have been sporadic and casual at best. These data are in close agreement with the diachronic analysis of the evidence for fire (indexed by burnt flints) at the Tabun cave site over a 300k year period: Shimelmitz et al. (2014) concluded that evidence for fire prior to c.350 ka is at best sporadic, but afterwards becomes regular and common, and that this pattern is repeated at a number of other Middle Pleistocene Levantine sites (including Qesem and Yabrud 1) (see also Shahack-Gross et al. 2014). It is notable that, even after 400 ka, not all populations made use of fire: only around 50% of occupation sites in Fig. 4 show evidence of fire. Sandgathe et al. (2011) present compelling evidence to suggest that fire was not used habitually by the Mousterian Neanderthal populations inhabiting southwestern France post-100 ka. This might account for the surprisingly large variance in cranial volume post-400 ka (Fig. 1).
As Wrangham (2010) has shown, cooking would have made it possible to acquire sufficient additional nutrient intake to make further large increases in brain volume possible without imposing additional demands on foraging time. With about 45% of the diet (of modern hunter-gatherers: Cordain et al. 2000) benefitting from a 50% increase in digestibility, there would be a net energy saving equivalent to about 45*0.5 = 22.5% of intake. On the de Miguel & Henneberg dataset, this would have allowed an increase in brain size from a typical cranial volume of ~ 1150cc just prior to 400 ka to one of ~ 1400 cc after 400 ka (equivalently, ~ 900 cc to ~ 1100 cc on the Pearce et al. dataset), roughly equivalent to the mean value achieved by archaic humans around 100 ka.
The timing for the regular cooking of food implied by the data in Figs. 1 is much later than that suggested by Wrangham (2010), who favoured an earlier date coinciding with the appearance of the genus Homo around 1.8 Ma. Wrangham (2010) offered three principal lines of evidence: a significant increase in brain size, a dramatic reduction in molar (chewing) tooth size and the first evidence for fire. We have already considered the third of these: first use is not the same as habitual use. An energetic benefit can only come from continuous use, not from sporadic use. What of the other two?
There is indeed an increase in cranial volume associated with the first appearance of the genus Homo around 1.8 Ma (Shultz et al. 2012). However, the increase in brain size at 1.8 Ma is quite modest compared to that associated with the appearance of archaic humans around 500 ka and, later, the Neanderthals and anatomically modern humans (Dunbar 2022). On average, H. ergaster brains were just 1.55 times the size of the average australopithecine brain (i.e. only half as big again), whereas those of Neanderthals and modern humans are 2.82 times as big (based on data from de Miguel & Henneberg 2001). If habitual cooking was necessary to fuel the brain growth associated with early Homo, then what was left to fuel the much larger increase among later humans?
If this represents the demand for cooking in modern humans, then the difference between the average australopith brain size and the average Homo ergaster brain (a ratio of 1:1.55) would represent the cooking demand for early Homo. Note, first, that cooking is really only relevant to the digestibility of roots, tubers and red meat, and in modern hunter-gatherers these foods account for only about 45% of the diet (Cordain et al. 2000). Given this, a 55% increase in energy intake could be achieved by cooking just 55/(0.45 * 0.50) = 12.4% of the diet (given that only 45% of the diet would yield the 50% increase in digestibility). This could certainly imply some level of cooking, but it would not be evidence for regular cooking. A hog roast once a month would probably suffice, and, at best, that would come under the category of opportunistic rather than habitual cooking. In fact, as noted earlier, an adjustment in the time required for social bonding (via the adoption of chorusing in the form of laughter) would easily have accounted for the increase in brain size in early Homo (Dunbar 2022). If the earlier increase in brain size can be accounted for without having to alter the diet, then we have the full value of the dietary explanation to account for the much larger increase in brain size that occurs in archaic humans. Cooking 45% of the diet would yield foraging savings of (45 * 0.5) * 182% = 41.0%, which would undoubtedly have a very significant impact on time budgets and additional foraging costs (Dunbar 2014a). This would be sufficient to allow the observed increase in cranial volume at around 400–500 ka.
Wrangham’s (2010) other line of evidence is the reduction in relative molar tooth size associated with early Homo. Although relative molar size in early Homo is certainly smaller than that in the australopithecines, in fact absolute molar size in early Homo is not that much smaller than that of the gracile australopithecines (Fig. 5). The apparently large change in relative molar size is mainly due to a dramatic increase in body mass rather than a reduction in molar size. The difference is crucial, because chewing efficiency is determined by absolute molar area, not by molar area relative to body mass. Since it seems that, despite their modest molar size, the gracile australopithecines were able to cope quite adequately with a high C4 diet (i.e. one based heavily on underground storage organs) (Sponheimer & Lee-Thorpe 2003; Sponheimer et al. 2005; Ungar & Sponheimer 2011) without the need for cooking, it is not obvious that early Homo would have been significantly worse off.
A possible fourth argument might be increased meat consumption in early Homo. It is certainly the case that early Homo ate meat, as is indicated by butchery sites, for example, at Olorgesailie dated to around 900 ka (Potts 1989). Whether this indicates more than just casual meat-eating of the kind already seen in the later australopithecines (McPherron et al. 2010) and modern chimpanzees (Fahy et al. 2013; Watts 2020) is open to debate. The important issue is that cooking does not make previously inedible meat digestible; it merely increases its digestibility by around 50% (Wrangham 2010). Like contemporary chimpanzees, early hominins could have eaten significant quantities of meat before it became necessary to use regular cooking to extract more nutrients. If the modest increase in brain size that came with H. ergaster needed only 12.4% of the diet to be cooked (see above), then ergaster could have solved its extra nutrient demand by a very modest 18.6% increase in its (uncooked) dietary intake (reflecting a difference in digestibility of 50% with cooking). The much larger nutrient demand for archaic and modern humans, on the other hand, could not have been more challenging.
Further indirect evidence in support of the current results comes from analyses of the gene for salivary amylase (AMY1), an enzyme that allows modern humans to convert glycemic carbohydrate starches into digestible glucose sugars. The efficiency with which amylase facilitates the digestion of starches increases three-fold after the starches have been gelatinized by cooking (Wang & Copeland 2013). This pathway is thought to be crucial in providing the energy supply needed to maintain our large brains (Hardy et al. 2015), and is especially important during fetal development and childhood (Bier et al. 1999; Butte 2000). The modern human genome contains more copies of the amylase gene than are found in other primates, including archaic humans. The presence of these extra gene copies has been interpreted as strong evidence for the habitual cooking of vegetable carbohydrates (Hardy et al. 2015). From a comparison of modern and archaic human genomes, Inchley et al. (2016) concluded that the likely date of origin of the modern human form is around 400–450 ka. If so, then the Neanderthals’ heavily meat-based diet would not have benefitted from this mechanism. Whether this is because the genetic calculations underestimate the true convergence dates or there was something unusual about the Neanderthal meat diet that obviated the need for this mechanism remains to be determined.
In sum, the anatomical and archaeological evidence provides support for Wrangham’s (2010) claim that cooking played an important role in hominin evolution, especially the impact it would have had on brain evolution. However, the data also suggest that cooking may have come on-stream a great deal later than Wrangham originally suggested, most likely around 400 ka. There is circumstantial evidence to suggest that this may have happened in the context of a significant thermally-driven energetic constraint immediately prior to ~ 400 ka that made it difficult to increase brain size (if not actually causing a decline in brain size as the climate deteriorated after 800 ka).