Temporal Dominance of Sensations: What is a good attribute list?

Temporal Dominance of Sensations (TDS) methodology recently gained more and more interest in the sensory field because of its unique approach enabling the description of the sequence of the dominant sensory perceptions along the tasting of a product. This method has been used for describing various products but it still faces some lack of knowledge on how to define a good list of attributes for this unusual sensory experiment. Indeed, TDS procedure is very different from standard sensory methodologies such as sensory profiling. Throughout the evaluation, panellists have to continuously make a choice among several attributes to determine the sequence of dominant sensations. Thus the definition of the attribute list is a key element since it determines the responses of the panellists. This paper investigates how panellists use the attribute list during TDS evaluation with a specific focus on the impact of the number, the type (texture, taste or aroma) and the position of attributes in the list on the panellist response and on the consensus among panellists regarding the use of this list. In order to get representative results, a database called “TDSbase” gathering TDS studies performed under different conditions was created in collaboration between CSGA (Dijon, France) and NRC (Lausanne, Switzerland). Results based on the 21 TDS studies from the TDSbase show that panellists tend to use a relatively constant number of attributes per evaluation, whatever the number of attributes in the list. Panellists are also able to use different types of attributes in the same list (no impact on the number of selections of each attribute or on the selection time). However, each panellist preferentially uses a subset of attributes that differs across panellists. In addition, if the list features more than 8–10 attributes, some panellists are not able to use them all. It is therefore recommended to use a list with a maximum of 10 attributes. Finally, the attribute order in the list does not impact the number of selection of each attribute, but the attributes at the top of the list tend to be selected earlier in the sequence than the attributes at the bottom of the list. It is therefore recommended to balance attribute order across panellists to account for this order effect.

► 21 TDS datasets were analysed to build knowledge on the use of the attribute list.
► Results provide guidance to panel leaders for the definition of the attribute list.
► The attribute list should not be too long (10 attributes maximum).
► The attribute order should be balanced across panellists to avoid a position bias.
► Several attribute types (e.g. texture and taste) can be used in the same list.