Sensory and Instrument-Measured Ground Chicken Meat Color

C. L. SANDUSKY1 and J. L. HEATH2 Department of Animal and Avian Sciences, University of Maryland, College Park, Maryland 20742

ABSTRACT Instrument values were compared to scores were compared using each of the backgrounds. sensory perception of ground breast and thigh meat The sensory panel did not detect differences in yellow- color. Different patty thicknesses (0.5, 1.5, and 2.0) and ness found by the instrument when samples on white background colors (white, pink, green, and gray), and pink backgrounds were compared to samples on previously found to cause differences in instrument- green and gray backgrounds. A majority of panelists (84 measured color, were used. Sensory descriptive analysis of 85) preferred samples on white or pink backgrounds. scores for lightness, hue, and chroma were compared to Red color of breast patties was associated with fresh- instrument-measured L* values, hue, and chroma. ness. Sensory ordinal rank scores for lightness, redness, and Reflective lighting was compared to transmission yellowness were compared to instrument-generated L*, lighting using patties of different thicknesses. Sensory a*, and b* values. Sensory descriptive analysis scores evaluation detected no differences in lightness due to and instrument values agreed in two of six comparisons breast patty thickness when reflective lighting was used. using breast and thigh patties. They agreed when thigh Increased thickness caused the patties to appear darker hue and chroma were measured. Sensory ordinal rank when transmission lighting was used. Decreased trans- scores were different from instrument color values in the mission lighting penetrating the sample made the patties ability to detect color changes caused by white, pink, appear more red. Reflective lighting made thigh patties green, and gray background colors. Instrument values appear lighter. Lightness decreased when thigh patty agreed with sensory scores for lightness only when thickness increased with both reflective and transmis- white and pink backgrounds were used. Instrument and sion lighting. Transmission lighting made the thigh sensory methods agreed when a* values and redness patties appear more yellow as patty thickness increased. (Key words: color, breast, thigh, sensory evaluation, sample thickness) 1998 Poultry Science 77:481–486

INTRODUCTION No research was found that compared sensory evaluation to instrument-measured ground chicken Ground chicken meat color is subject to critical meat color. Most of the ground meat color research has appraisal in the processing plant and is important in been done using red meat and pork. Research on retail sales. The effect of reflectance and transmission ground meat from other species indicated that color is lighting on product appearance are important in deter- perhaps the most important influence in consumer mining product display and marketing strategies. Sen- decision-making regarding acceptability of fresh meat sory evaluation is the best method of measuring products for consumption (Brewer and Harbers, 1991; consumer response to product color but it is slow and Hunt et al., 1993). Lynch et al. (1986) reported that 74% of requires a large investment in people and facilities. consumers indicated that color was important in ground Efficient, cost-effective, and highly sensitive instrumen- meat purchase decisions. Mugler and Cunningham tation is available to measure color and is often used (1972) reviewed factors affecting poultry meat color, instead of sensory measurement. Sensory perception of including sex, age, strain, processing procedures, chemi- color is multidimensional and may be difficult to cals, cooking, irradiation, and freezing. Patty thickness, measure with an instrument. Instrument-measured color background color, and tissue type also affect instrument must accurately predict sensory response to be useful in measured ground chicken meat color (Sandusky and most quality and all consumer preference evaluations. Heath, 1996). The first objective of the present studies was to determine whether instrument-measured color of Received for publication June 5, 1997. ground chicken meat would predict color differences Accepted for publication October 20, 1997. perceived by a sensory panel. The second objective was 1Present address: Protein Technologies International, Annapolis, to compare the effect of reflective and transmission MD 21401. 2 To whom correspondence should be addressed: lighting on sensory evaluation of ground chicken meat [email protected] color.

481 482 SANDUSKY AND HEATH MATERIALS AND METHODS Descriptive analysis was chosen as one sensory method used to measure patty color. This method has been used to Broiler chicken breast and thigh meat were obtained determine how specific product differences are related to from a commercial processing plant immediately after differences in instrumental measures (Stone et al., 1974). A deboning. Fresh breast and thigh meat were obtained for 15-cm line scale (Anderson, 1970) with two descriptive each study and each replication within a study. The anchors as described in the spectrum descriptive proce- Pectoralis superficialis, Illiotibialis, and Quadraceps femoris dure (Meilgaard et al. 1991) was used. A middle anchor muscles were used to prepare breast and thigh meat was not used on the line scale because its use has been patties as described by Sandusky and Heath (1996). The reported to increase variability in panel response by 10 to ground tissue was mixed thoroughly and the patties 15% (Stone and Sidel, 1985). Panelists were not told that were randomly assigned to treatments to randomize the scale had numerical context because this could induce possible differences due to preparation. bias (Stone and Sidel, 1985). Panelists were asked to make a vertical line across the horizontal line scale at the point Instrument Color Measurement that best represented the relative intensity or hue of a particular sample. The distance of the panelist’s mark Color measurements were made with a Model CS-53 from the left anchor was measured and converted to dual beam scanning spectrophotometer as described by percentage of line length to obtain a numerical value for Sandusky and Heath (1996). Instrument calculated repeat- statistical analysis. Red and yellow anchors were used to ± ± ability and accuracy of data were 0.05 and 0.3 nm, measure hue. Previous work (Sandusky and Heath, 1996) respectively. Specular reflectance was included in the showed that instrument measured hue angles for ground instruments program and added approximately 4% to the chicken meat were in the quadrant bordered by the red percentage reflectance of the sample at any wavelength. and yellow axes of the color space model. Light and dark Illuminant F representing three band fluorescent with a anchors were used to measure lightness and dull and color temperature of 4,200 K was used as the light source. bright anchors were used to measure chroma. Hunter4 standard white (6550), pink (6551), green (6552), Ordinal ranking was another sensory method used to and gray (6555) tiles were used to provide background measure color. Ordinal ranking measures perceived color for the samples during measurement. The CIELAB intensities of product color and works well within the color space model (CIE, 1978) was chosen to numerically cognitive limits of most panelists. Samples were ranked describe color. L* (lightness) describes the relationship from least to most using the numbers 1 to 4 (Meilgaard et between reflected and absorbed light, without regard to al., 1991). Patty lightness, redness, and yellowness were specific wavelength. Positive a* values are red and evaluated using this procedure. In addition to ranking the negative a* values are green. Positive b* values are yellow samples, panelists were asked which sample they and negative b* values are blue. Chroma is a measure of preferred and they were asked to write descriptive color saturation and hue is the color angle (CIE, 1978). comments on the evaluation form. Samples were evalu- ated by the panelists in a specific order randomly selected Sensory Panel Evaluation by the researchers.

Panelists were required to demonstrate normal color Study 1 vision as suggested by Billmeyer and Saltzman (1981). Normal color vision was defined as the ability to Sensory measurement of hue, lightness, and chroma discriminate between red, green, blue, and yellow colors using descriptive color analysis was compared to and the ability to sort colored samples into a regular series instrument-measured hue, L*, and chroma. Patty thick- involving a gradual change in hue. Panelist were trained nesses of 0.5, 1.5, and 2.0 cm were used to provide breast in the Munsell hue, value, and chroma order system and thigh samples that produced statistically different (Munsell, 1963) to reduce observer variability within a instrument values in previous research (Sandusky and group with normal color vision (Billmeyer and Saltzman, Heath, 1996). Each of the 10 to 12 panelist evaluated the 1980). Panelists were given a verbal explanation of the color of each patty (three thicknesses) three times for a definition of hue, value, and chroma and a Munsell total of nine observations per panelist for each tissue type teaching chart was used to help panelists visually (breast and thigh). Breast and thigh patties were tested differentiate color (hue), shades (value), and intensity separately to avoid color bias. Patties were placed on a (chroma). After the panelists displayed a clear level of white background and lighted from above with two 15 W confidence with these terms, they were presented with cool white fluorescent bulbs (4,200 K) that were 0.75 m actual samples in the sensory booth. Equal numbers of above and at a 90° angle to the sample to provide reflective male and female panelists, whose age ranged from 20 to lighting. Panelists were instructed to place their heads at a 50+ yr, formed the pool of panelists. specific measured location to ensure that samples were always viewed from the same height and angle. Each sample evaluated by the sensory panel was also measured 3Model CS-5, Applied Color Systems, Princeton, NJ 08540. with the CS-5 spectrophotometer using the same white 4Hunter Associates Laboratory, Inc., Fairfax, VA 22030. background. The study was replicated twice with an INSTRUMENT AND SENSORY COLOR COMPARISON 483 average of 32 evaluations for each patty thickness and values in Study 2. Data from sensory analyses were tissue type in each replication. analyzed by the nonparametric Rank Sum Procedure and the Friedman statistic was used to test for significance (P < Study 2 0.05) (Civille, 1992). Statistical analysis of the instrument data was done using the Randomized Complete Block Sensory evaluation of breast meat patties on different design procedure and means were separated using background colors was compared to instrument measure- Duncan’s multiple range test (SAS Institute, 1985). ment of the same samples. The ordinal ranking procedure Correlation analysis was used to determine the joint was chosen because it is not as variable as descriptive relationships between instrument values and sensory sensory analysis and is simple and repeatable (Meilgaard scores (SAS Institute, 1985). Instrument values and et al., 1991). Patties (1.0-cm thick) were placed on white, sensory scores were converted to percentages to facilitate pink, green, or gray backgrounds and lighted from above graphic comparison of values and scores that differed as described in Study 1. The white, pink, green, and gray substantially in magnitude. In Study 3, treatments for both transmission and backgrounds were found to cause significant differences reflection lighting studies were arranged in a 2× 3 factorial in instrument values in previous experiments (Sandusky with two types of meat (breast and thigh) and three patty and Heath, 1996). Four-centimeter diameter circles were thicknesses (0.5, 1.0, and 1.5 cm). ANOVA (P < 0.05) and removed from white polystyrene sheets and placed correlation were used to statistically analyze the data (SAS around the patties to cover the exposed portion of the Institute, 1985). background tiles. The 10 to 12 panelists were instructed to place their heads at a specific measured location during the evaluation. Each panelist ranked samples placed on RESULTS AND DISCUSSION each background tile and the evaluation was replicated (n = 20). The CS-5 spectrophotometer was used to measure Study 1 each sample evaluated by the sensory panel on the same Instrument-measured and sensory perception of patty background tiles. Instrument color was measured as color were different in six of eight comparisons (Table 1). described previously. Instrument L* values showed lower lightness values when breast patty thickness was increased from 0.5 to 1.5 cm. Study 3 The lower lightness values were attributed to increased patty thickness reducing the amount of light reflected Sensory color scores from patties illuminated with from the background and agreed with results reported by reflection and transmission lighting were compared. The Sandusky and Heath (1996). The sensory panel did not descriptive analysis procedure described in Study 1 was detect this decrease in breast patty lightness. The used. Transmission lighting was provided by a 5,000 K instrument did not detect the increase in thigh patty fluorescent source beneath the sample. Three samples lightness (L* values) found by the sensory panel when were presented in a horizontal row on a 33 × 43 cm light patty thickness was increased from 0.5 to 1.5 cm. board tilted at a 30° angle inside a white sensory analysis Apparently, the sensory perception of lightness in breast booth. The samples received some light from normal and thigh patties included factors other than the fluorescent room lighting in addition to the transmission instrument-measured amount and wavelengths of light lighting. Reflection lighting was tested using samples, reflected from the samples or background. booth area, and light board as described for transmission Instrument-measured hue increased when breast patty testing except the light board did not provide transmis- thickness increased from 0.5 to 1.5 cm, indicating a change sion lighting. A white opaque cover was placed on top of in color away from red toward a more yellow color. The the light board to serve as background for the samples. hue angle moved counter clockwise from the + a* (red) Reflectance lighting was provided by two 15 W cool white axis toward the + b* (yellow) axis on the CIE color space fluorescent bulbs (4,200 K) located 0.75 m above the light model. The sensory panel using a scale anchored by red board. Hue, lightness, and chroma were measured as and yellow did not detect this change in hue. No other hue value differences were detected by the sensory panel or described in Study 1. The comparison was replicated and instrument. 10 to 12 panelists were used in each trial (n = 22). Instrument-measured breast patty chroma (color inten- sity) decreased when patty thickness was increased from Statistical Analysis 0.5 to 1.5. This decrease was not found by the sensory panel using a scale anchored by dull and bright. Both the An ANOVA was used in Study 1 to test for significant instrument and sensory panel found a significant decrease (P < 0.05) differences between sensory and instrument in chroma when thigh patty thickness was increased from measurements of hue, lightness (L*), and chroma. Signifi- 0.5 to 1.5. The sensory panel found a further decrease in cantly different means were separated using Duncan’s thigh patty chroma when thickness was increased from 1.5 multiple range test (SAS Institute, 1985). to 2.0 cm that was not detected by the instrument. Sensory data for lightness, redness, and yellowness Care should be taken when using instrument values to were compared to instrument measured L*, a*, and b* predict statistical differences in sensory response to 484 SANDUSKY AND HEATH

FIGURE 1. Comparison of instrument measured L* values to sensory FIGURE 2. Comparison of instrument measured a* values to sensory lightness scores from ground breast meat on white, pink, green, and gray redness scores from ground breast meat on white, pink, green, and gray backgrounds using the ordinal ranking system. Means for each backgrounds using the ordinal ranking system. Means for each measurement (sensory or instrument) with different letters differ measurement (sensory or instrument) with no common letter (a,b and significantly (P < 0.05). Sensory scores and instrument values were x,y, respectively) differ significantly (P < 0.05). Sensory scores and converted to percentages to facilitate plotting. instrument values were converted to percentages to facilitate plotting.

ground chicken meat color. This work differs from other The sensory panel did not detect the differences in work in which descriptive analysis was found to be yellowness found by the instrument (Figure 3). The especially helpful in identifying product differences instrument found greater b* values for samples on the related to differences in instrumental values (Stone et al., white and pink backgrounds than for samples measured 1974). on the green and gray backgrounds. Study 1 and this study indicated that the instrument often failed to accurately predict differences in sensory Study 2

Lack of agreement between instrument values and sensory descriptive analysis scores in Study 1 prompted the use of the ordinal ranking procedure. Instrument produced L*, a*, and b* values were compared to sensory lightness, redness, and yellowness scores produced by the ordinal ranking procedure. The sensory panel ranked samples on white and pink backgrounds as lighter than samples on green and gray backgrounds (Figure 1). Lightness scores obtained using white and pink backgrounds were not statistically different from each other. The green background produced higher lightness scores than the gray back- ground. The instrument did not detect any differences in L* values (lightness) when the same four background tiles were used. The instrument agreed with sensory evalua- tion of lightness only in the lack of difference between samples on white and pink backgrounds. Instrument-measured a* values and sensory scores for redness detected the same changes in patty color (Figure FIGURE 3. Comparison of instrument measured b* values to sensory 2). The a* values and redness scores were greater for yellowness scores from ground breast meat on white, pink, green, and samples measured with white and pink backgrounds than gray backgrounds using the ordinal ranking system. Means for each measurement with no common letter differ significantly (P < 0.05). for those measured using the green and gray back- Sensory scores and instrument values were converted to percentages to grounds. facilitate plotting. INSTRUMENT AND SENSORY COLOR COMPARISON 485

FIGURE 4. Sensory descriptive analyses of ground breast meat using FIGURE 5. Sensory descriptive analyses of ground thigh meat using reflection (R) and transmission (T) lighting. Bars with the same letter for reflection (R) and transmission (T) lighting. Bars with the same letter for each lighting type and sensory measure are not significantly (P > 0.05) each lighting type and sensory measure are not significantly (P > 0.05) different. Lightness, hue, and chroma were measured using light and different. Lightness, hue, and chroma were measured using light and dark, red and yellow, and dull and bright anchors, respectively. dark, red and yellow, and dull and bright anchors, respectively. scores when the respective measures were analyzed statistically significant differences by the sensory panel statistically. Examination of the data indicated that and the instrument could be due to sensory panel instrument and sensory measures demonstrated similar preference and other visual stimuli rather than a narrow trends. Based on this observation, the joint relationship perception of differences in lightness, redness, or yellow- between instrument values and sensory scores was ness. measured. Correlation coefficients were calculated using Panelists were asked to identify which patties they data from 1.0-cm-thick samples measured on the white would most like to cook and eat. The panelists preferred background. Correlation coefficients for L* values and (84 of 85) patties on white or pink backgrounds over those lightness scores and for a* values and redness scores were presented on green or gray backgrounds. Sensory scores highly significant (P < 0.01), r = 0.93 (Y = –3365.25 + 56.17x) from the ordinal ranking procedure were larger for and 0.99 (Y = –25.95 + 17.09x), respectively. Correlation samples on the white and pink backgrounds than for coefficients for b* values and yellowness were significant samples viewed on the green or gray backgrounds when (P < 0.05), r = 0.77 (Y = 48.83 + 6.9x). These data indicate lightness (Figure 1) and redness (Figure 2) were measured. that the instrument values and sensory panel scores The instrument detected larger values for the white and measured the same trends, as indicated by their linear pink backgrounds when a* (redness, Figure 2) and b* relationships. The difference in ability to detect the same (yellowness, Figure 3) values were analyzed.

TABLE 1. Color measurement of ground chicken meat patties using instrument measurement and descriptive sensory analysis

Patty thickness Color Color measurement Tissuevalue 0.5 cm 1.5 cm 2.0 cm SEM (%) Sensory Breast Lightness1 62.4a 54.8a 71.2a 4.5 Thigh 92.2a 56.6b 70.2b 5.6 Instrument Breast L* 65.4a 61.1b 60.2b 2.7 Thigh 60.9a 59.9a 59.4a 2.9 Sensory Breast Hue2 83.7a 100.0a 95.9a 6.7 Thigh 39.6a 55.1a 56.1a 4.2 Instrument Breast Hue 68.2b 70.8a 71.4a 2.9 Thigh 69.5a 70.4a 70.4a 3.1 Sensory Breast Chroma3 86.2a 74.0a 75.1a 5.1 Thigh 112.1a 81.9b 41.7c 5.6 Instrument Breast Chroma 21.0a 19.0b 17.5b 0.8 Thigh 19.5a 17.7b 16.3b 0.9 a–cMeans in rows with no common superscript differ significantly (P < 0.05). 1Light and dark anchors were used on the line scale. The darker sample has the larger score. 2Red and yellow anchors were used on the line scale. 3Dull and bright anchors were used on the line scale. 486 SANDUSKY AND HEATH Written comments by panelists indicated they as- Correlation coefficients were calculated using sensory sociated red color of ground chicken breast meat with scores for each of the patty thicknesses to compare freshness. The instrument values indicated that the white transmission to reflection lighting. Significant correlations and pink backgrounds reflected more light in the red (P < 0.05) were found for thigh meat hue (r = 0.83, Y = 12.45 wavelengths, which explained the preference by panelists + 1.11x), breast meat hue (r = 0.78, Y = 318.40 – 2.59x), thigh for samples on these two backgrounds. Lower scores for meat chroma (r = 0.99, Y = 47.49 + 0.33x), breast meat patties evaluated on the green and gray backgrounds chroma (r = 0.80, Y = 66.82 + 0.15x), and thigh meat compared to the white and pink backgrounds may explain lightness (r = 0.61, Y =145.0 – 0.97x). Lightness values for the panelists’ comments that the green and gray back- breast meat indicated there was no correlation between grounds made the samples appear darker, older, and less transmission and reflection lighting (r = 0.45). fresh. REFERENCES

Study 3 Anderson, N. H., 1970. Functional measurement and psychologi- cal judgement. Psych. Rev. 77:153–170. Point of sale displays of ground chicken products often Billmeyer, F. W., and M. Saltzman, 1980. Observer metamerism. use light transmitted through the product from below in Color Res. Appl. 5:72. addition to light reflected from the product surface. Billmeyer, F. W., and M. Saltzman, 1981. Principles of Color Transmission lighting is accomplished by packaging in Technology. 2nd ed. John Wiley and Sons, New York, NY. clear plastic trays and using light beneath the product. Brewer, M. S., and C.A.Z. Harbers, 1991. Effect of packaging on Studies 1 and 2 indicated that differences in the amount of color and physical characteristics of ground pork in long light reflected by the sample and background measured term frozen storage. J. Food Sci. 56:363–366, 370. with an instrument often did not affect sensory evalua- C.I.E., 1978. International Commission on Illumination, Recom- mendations on uniform color spaces, color-difference tion. In this study, sensory evaluation was used to equations, psychometric color terms. Supplement no. 2 to determine differences in color perception when reflection C.I.E. publication no. 15(E-1.31) 1971/(TC-1.3) 1978. Bureau and transmission lighting was used and patty thickness de la C.I.E., Paris, France. increased. Civille, R., 1992. Sensory Evaluation. Academic Press, New York, No differences due to breast patty thickness were found NY. for lightness, hue, and chroma when reflection lighting Hunt, M. C., D. H. Kropf, and J. B. Morgan, 1993. Color was used (Figure 4). Transmission lighting made the measurement of meat and meat products. Pages 59–60 in: patties appear darker when patty thickness was increased Proceedings, 46th Reciprocal Meat Conference. American Meat Science Association, Chicago, IL. because less light penetrated the patties from below. Also, Lynch, N. M., C. L. Kastner, and D. H. Kropf, 1986. Consumer less transmission light passed through the sample to the acceptance of vacuum packaged ground beef as influenced observer as patty thickness was increased, which resulted by product color and educational materials. J. Food Sci. 51: in the perception of decreased hue (redness). Increased 253–255, 272. redness was perceived by the sensory panel to be an Meilgaard, M., G. V. Civille, and B. T. Carr, 1991. Sensory indication of freshness in Study 2. Transmission and Evaluation Techniques. 2nd ed. CRC Press, Inc., Boca Raton, reflection lighting resulted in no change in breast tissue FL. chroma (brightness) as patty thickness increased. Mugler, D. J., and F. E. Cunningham, 1972. Factors affecting poultry meat color—A review. World’s Poult. Sci. J. 28: An increase in lightness was found when thigh patty 400–406. thickness was increased and reflection type lighting used Munsell, A. H., 1963. A Color Notation. Munsell Color Co., (Figure 5). Lightness was decreased as thigh patty Baltimore, MD. thickness increased and transmission lighting was used. Sandusky, C. L., and J. L. Heath, 1996. Effect of background Thigh patties appeared more yellow (increased hue) as color, sample thickness, and illuminant on the measurement patty thickness increased and transmission lighting was of broiler meat color. Poultry Sci. 75:1437–1442.  used. No change in hue was found when reflection SAS Institute, 1985. SAS/STAT Guide for Personal Computers. lighting was used. Both transmission and reflection Version 6. SAS Institute Inc., Cary, NC. Stone H., and J. L. Sidel, 1985. Sensory Evaluation Practices. lighting resulted in decreased brightness as patty thick- Academic Press, Inc., New York, NY. ness increased. Choice of lighting type will depend on Stone, H., J. Sidel, S. Oliver, A. Woolsey, and R. C. Singleton, whether light or dark meat patties are displayed based on 1974. Sensory evaluation by quantitative descriptive analy- sensory panel responses. sis. Food Technol. 28:24–34.