Body is one of the fundamental terms used in describing coffee. The SCA defines it as “the tactile feeling of coffee in the mouth,” a description broad enough to include many different phenomena.

JESPER ALSTRUP explores “body” and how roast profile modulations affect the sensory experience of brewed coffee through his research at the University of Copenhagen, supported in part by the Specialty Coffee Association, in Issue 5 of 25 Magazine.

After meeting and working with the coffee community, I became fascinated with the many ways coffee professionals used the term body and the techniques roasters said they used to manipulate it. I saw coffee experts struggle to reach consensus on the meaning of body, even while they suggested techniques of developing it in the final product. These observations and questions moved me to devote my master’s thesis to the concept of body, its sensory perception, and the effect of various modulations of the roasting process on the perception of this important, and sometimes elusive, attribute.

What is “body,” anyway?

It was imperative to begin by exploring the meanings of body before deciding on our approach to measuring it: though SCA’s definition of body is widely accepted, a broad range of interpretations can be found in practice. Through the use of an anonymous questionnaire and in-depth expert interviews including more than 70 coffee professionals, we found an amazingly wide variety of specific descriptors tasters use to describe body: viscous, coating, heavy, syrupy, astringent, etc. In all, we noted 34 different individual descriptive words relating to the concept of body. We also found two common approaches to evaluating “body” in coffee: those who evaluate body based on intensity and quality (both positive and negative attributes) and those who evaluate on intensity alone.

Finally, we asked these coffee experts about the cause of high body in coffee, and many mentioned roast degree and overall roast development time as determining factors.

It was during this time that we narrowed our focus and developed our hypotheses. A key requirement for the study was to use techniques that are broadly relevant to the industry, so we kept our focus on the two determining factors mentioned by our experts: roast degree, as measured by color, and roast development, or the length of time between first crack and the end of the roast. We discovered many roasters had beliefs about modulating body through roasting. Particularly, there is a common belief that a coffee body is at its lightest at light roasts, increasing in “intensity” as the roast evolves. After reaching a peak at medium roast, body begins to lighten again, creating a kind of “hump” of optimal coffee heaviness. We also heard from experts that body could be increased by extending the development time of the roast – the time between first crack and the end of the roast – without changing the final roast degree as measured by color. In other words, extending the roast development time was a method to increase body.

Under the supervision of Morten Münchow, who serves as SCA’s research liaison and advisor at the University of Copenhagen, we set out to design a set of experiments to determine how different roast levels and profiles might change the sensory perception of brewed coffee. Although we tracked 10 total sensory attributes throughout the study (overall aroma intensity, acidity, bitterness, sweetness, grittiness, astringency, roasted flavor, dried fruit flavor, nutty/chocolate flavor, and body), the primary focus remained on the sensory descriptor “body” and how it modulates through a variety of roast profiles.

The Experiments

Fig. 1 Overview of the relationship between the roast profiles in the study. The medium sample serves as the anchor between the two types of modulations, having a medium roast degree of Agtron 75 and a development time of 2 min. 15 sec.

We began by designing the roast profiles we would use to explore our hypotheses, and we ended up with a selection of nine roasts: lightest, light, medium, dark, darkest, fast, slow, baked, and baked+. (Fig. 1). These final nine profiles needed to be perfectly repeatable in order to ensure useable data. For roast degree samples, only the final color of the coffee was adjusted – this meant keeping a development time (the time between first crack and the end of roast) of around 2 min. 15 sec. for all samples in this category. For roast development samples, only the development time was adjusted – this meant keeping the roast color of all samples consistent (Agtron 75, or medium roast).

By focusing on one modulation at a time, any variation in flavor profile would be due to the specific changes made during the roast. All roast data was logged using the Cropster roasting software, from which we could extract the roast data for all the samples.

Sensory Descriptive Analysis

How did these changes influence the flavor of the brewed coffee? This can only be scientifically answered through descriptive sensory analysis (DSA). Although it appears similar to a traditional coffee tasting session, DSA is a much different process. First, DSA uses a professional sensory panel, trained in the rigorous technique of descriptive analysis – for the purpose of this study, I used a panel of coffee professionals who had been trained in the descriptive analysis process. The panel used some commonly used references for taste and aroma, but we also provided special references to help them define concepts associated with body: skim and full-fat milks serving as references for thickness and weight, a flour-water solution to serve as a reference for grittiness, over-extracted tea as a reference for astringency, and various solutions of xanthan gum (a food thickener) and water to serve as a reference for viscosity. References like these are essential for the DSA process to work effectively, and help panelists precisely rate specific sensory descriptors in coffee. Our panel of 10 participants evaluated each sample in a private, blind setting, completely unaware of which coffee they received. From their isolated “tasting booth”, working in complete silence, each panelist was asked to rate the intensity of the sensations they perceived on a continuous 15 cm line scale. It was only once we had compiled the results of the DSA evaluations and tested them for statistical significance that we could begin to interpret the data.

The Results

Roast Degree Modulations (Fig. 2)

Fig. 2 Overview of changes to the flavor profile by adjusting roast degree. Other attributes were included in the study, such as sweetness, but they were not statistically significantly different.

What happens to the flavor profile when we adjust the roast degree but keep roast time the same? Here is what we found: in the lightest roast (Agtron 98) the panel described high “acidity” and “dried fruit” flavors with a lower perception of “bitterness” and “roasted.” Reducing the Agtron number (i.e. roasting darker) led to increased perceptions of “bitter,” “gritty,” and “roasted” attributes, while reducing the perceived “acidity” and “fruitiness.” Turning our attention to body, we see an interesting phenomenon: “body” ratings begin low, but increase as the roast evolves. Later, as roasts become darker, the body appears to lighten again – suggesting that body does not have a linear relationship with roast degree. However, this decline in body did not prove to be statistically significant in this study.

Fig. 3 Overview of changes to the flavor profile by adjusting development time. Other attributes were included in the study, such as sweetness and bitterness, but they were not statistically significantly different.

Development Time Modulations (Fig. 3)

We then turned our attention to our other hypothesis that modulating roast development time – the duration of the roast between first crack and the end of the roast – could help us modulate body scores. We had developed some pretty extreme roast profiles – our shortest roast development time was 1 min. 15 sec., and our longest was 6 min. 32 sec. All samples were of the same color, Agtron 75. We found much smaller differences in the sensory perception of each sample. Most interestingly, body was not perceived as much different throughout the development time modulations. Thus, our results disproved our initial hypothesis that “body” would increase with roast development time.

Viscosity Measurement (Fig. 4)

Fig. 4 Measurements of viscosity. Greatest viscosity differences exist with differences in roast degree.

I wanted to include an instrumental measurement of “viscosity,” a term often used by coffee professionals as a synonym for body. We measured the viscosity of the various coffees using an instrument called a viscometer, which measures the energy needed to vibrate two plates, exactly 1 mm apart, with the coffee in between them. This makes viscosity a good, standard, physical measurement of coffee thickness, and perhaps an indicator of potential body perception. We saw a small yet clear difference in viscosities among the various coffees, and viscosity differences were particularly evident in variations between different roast levels, the darkest coffee having by far the lowest viscosity of all the samples.

The viscosity measurements revealed an interesting reverse u-shaped trend for roast degree modulations, similar to the trends of body development, with body and viscosity both peaking at Agtron 75. However, this evidence is not sufficient proof of a link between viscosity measurements and the sensory perception of body – more research is required.

Why Is This Important?

Some of the results of the study may seem to be obvious, but the implications go far beyond the results above. In order for us to map the sensory effects of roast modulations, we first had to propose a measurable definition of body, so in effect our study also functioned as a test of this new definition. Was it possible to define body using measurable terms and produce statistically significant results?

The results of this study confirm that body is indeed a measurable concept, when defined as “viscosity, heaviness, and coating nature of the beverage” and evaluated solely for intensity. As most specialty coffee professionals involved were also inclined to evaluate body for quality, we found that introducing a separate attribute scale (smooth/pleasant to gritty/astringent) was necessary to cover the mouthfeel sensations that would otherwise be included in qualitative evaluations of body.

An early hypothesis of this study, built from expert opinion, personal experience, and bias (including our own) that body could be perceptibly increased as development time increased, was shown to be statistically insignificant. While these may not be the exact results for which we had hoped, there is perhaps an even better outcome: proof that research is necessary to challenge what we think we know, so that we can move forward with greater clarity.

There are obviously other more complicated roast modulations possible that are likely to affect the final flavor profile in brewed coffee, but it is important to map the effects of small changes at a very basic, foundational level before exploring additional, more complex modulations. One thing remains true regardless: continued scientific research is the only way to determine whether established community “truths” are scientifically repeatable, or whether the reality of coffee is more elusive, complex, and interesting.

JESPER ALSTRUP is a sensory scientist at CoffeeMind (Copenhagen, Denmark). He received his M.Sc. in Food Innovation and Health from the University of Copenhagen. Support for this research comes from the members of the Specialty Coffee Association.

Modulating Perception - 25 Magazine: Issue 5